src/pages/potential_analysis.py

import streamlit as st
import sys
import os
import plotly.graph_objects as go
import matplotlib.pyplot as plt
import tempfile
import csv
from io import BytesIO
from PIL import Image, ImageDraw, ImageFont
from graphics_utils import draw_smooth_ellipse
from datetime import datetime

# # Add the src directory to the path for imports
# src_path = os.path.abspath(os.path.join(os.path.dirname(__file__), '..'))
# if src_path not in sys.path:
#     sys.path.insert(0, src_path)

# Import pinch analysis modules
try:
    from Modules.Pinch.Pinch import Pinch
    from Modules.HeatPumpIntegration.HeatPumpIntegration import HeatPumpIntegration as HPI
    from Modules.Pinch_main import Pinchmain
    PINCH_AVAILABLE = True
    PINCH_IMPORT_ERROR = None
except ImportError as e:
    PINCH_AVAILABLE = False
    PINCH_IMPORT_ERROR = str(e)

st.set_page_config(
    page_title="Potential Analysis",
    initial_sidebar_state="collapsed"
)

# Apply styles immediately to prevent flash
st.markdown(
    """
    <style>
    :root {
        font-size: 11px !important;
    }
    section[data-testid="stSidebar"][aria-expanded="true"] {
        width: 180px !important;
        min-width: 180px !important;
    }
    section[data-testid="stSidebar"][aria-expanded="false"] {
        width: 0 !important;
        min-width: 0 !important;
        margin-left: 0 !important;
    }
    
    /* Smaller fonts and elements - apply to all elements */
    html, body, .stApp, * {font-size:11px !important;}
    .stMarkdown p, .stMarkdown span, .stMarkdown li {font-size:11px !important; margin:0 !important; padding:0 !important;}
    .stButton button {font-size:10px !important; padding:0.1rem 0.3rem !important;}
    .stTextInput input, .stNumberInput input {font-size:10px !important; padding:0.1rem 0.2rem !important;}
    h1 {font-size: 1.5rem !important; margin-bottom: 0.3rem !important;}
    /* Compact layout */
    .block-container {padding-top: 2rem !important; padding-bottom: 0 !important;}
    div[data-testid="stVerticalBlock"] > div {padding: 0 !important; margin: 0 !important;}
    hr {margin: 0.3rem 0 !important;}
    .stCheckbox {margin: 0 !important; padding: 0 !important;}
    div[data-testid="stHorizontalBlock"] {gap: 0.2rem !important;}
    </style>
    """,
    unsafe_allow_html=True,
)

# Helper function to convert lon/lat to pixel coordinates (same as data_collection.py)
def snapshot_lonlat_to_pixel(lon, lat, center_ll, z_level, img_w, img_h):
    import math as _math
    def lonlat_to_xy(lon_val, lat_val, z_val):
        lat_rad = _math.radians(lat_val)
        n_val = 2.0 ** z_val
        xtile = (lon_val + 180.0) / 360.0 * n_val
        ytile = (1.0 - _math.log(_math.tan(lat_rad) + 1 / _math.cos(lat_rad)) / _math.pi) / 2.0 * n_val
        return xtile, ytile
    
    center_lon, center_lat = center_ll
    cx, cy = lonlat_to_xy(center_lon, center_lat, z_level)
    px_tile, py_tile = lonlat_to_xy(lon, lat, z_level)
    tile_size = 256
    dx_tiles = px_tile - cx
    dy_tiles = py_tile - cy
    dx_px = dx_tiles * tile_size
    dy_px = dy_tiles * tile_size
    screen_x = img_w / 2 + dx_px
    screen_y = img_h / 2 + dy_px
    return screen_x, screen_y

def generate_process_level_map():
    """Generate a map image showing only the process-level (green boxes)."""
    snapshots_dict = st.session_state.get('map_snapshots', {})
    active_base = st.session_state.get('current_base', 'OpenStreetMap')
    
    if active_base == 'Blank':
        base_img = Image.new('RGBA', (800, 600), (242, 242, 243, 255))
    else:
        chosen_bytes = snapshots_dict.get(active_base) or st.session_state.get('map_snapshot')
        if not chosen_bytes:
            return None
        base_img = Image.open(BytesIO(chosen_bytes)).convert("RGBA")
    
    w, h = base_img.size
    draw = ImageDraw.Draw(base_img)
    
    # Load font
    BOX_FONT_SIZE = 20
    try:
        font = ImageFont.truetype("/System/Library/Fonts/Arial.ttf", BOX_FONT_SIZE)
    except (OSError, IOError):
        try:
            font = ImageFont.truetype("/System/Library/Fonts/Helvetica.ttc", BOX_FONT_SIZE)
        except (OSError, IOError):
            try:
                font = ImageFont.truetype("DejaVuSans.ttf", BOX_FONT_SIZE)
            except (OSError, IOError):
                font = ImageFont.load_default()
    
    # Draw process-level boxes (green)
    group_coords = st.session_state.get('proc_group_coordinates', {})
    proc_group_names = st.session_state.get('proc_group_names', [])
    map_center = st.session_state.get('map_center', [0, 0])
    map_zoom = st.session_state.get('map_zoom', 17.5)
    
    for group_idx, coords_data in group_coords.items():
        group_idx = int(group_idx) if isinstance(group_idx, str) else group_idx
        if group_idx < len(proc_group_names):
            lat = coords_data.get('lat')
            lon = coords_data.get('lon')
            if lat is not None and lon is not None:
                try:
                    lat_f = float(lat)
                    lon_f = float(lon)
                    group_px, group_py = snapshot_lonlat_to_pixel(
                        lon_f, lat_f,
                        (map_center[1], map_center[0]),
                        map_zoom, w, h
                    )
                    if group_px < -50 or group_py < -20 or group_px > w + 50 or group_py > h + 20:
                        continue
                    
                    group_label = proc_group_names[group_idx]
                    scale = float(coords_data.get('box_scale', 1.5) or 1.5)
                    padding = int(8 * scale)
                    text_bbox = draw.textbbox((0, 0), group_label, font=font)
                    tw = text_bbox[2] - text_bbox[0]
                    th = text_bbox[3] - text_bbox[1]
                    box_w = int(tw * scale + padding * 2)
                    box_h = int(th * scale + padding * 2)
                    x0 = int(group_px - box_w / 2)
                    y0 = int(group_py - box_h / 2)
                    x1 = x0 + box_w
                    y1 = y0 + box_h
                    
                    # Draw box
                    fill_color = (200, 255, 200, 245)
                    border_color = (34, 139, 34, 255)
                    text_color = (0, 100, 0, 255)
                    draw.rectangle([x0, y0, x1, y1], fill=fill_color, outline=border_color, width=3)
                    
                    # Draw label
                    text_x = x0 + (box_w - tw) // 2
                    text_y = y0 + (box_h - th) // 2
                    draw.text((text_x, text_y), group_label, fill=text_color, font=font)
                    
                    # Draw stream circles above the process box
                    group_subprocess_list = st.session_state.get('proc_groups', [])[group_idx] if group_idx < len(st.session_state.get('proc_groups', [])) else []
                    all_streams = []
                    selected_items = st.session_state.get('selected_items', {})
                    
                    # Collect selected streams from all subprocesses
                    for sub_idx in group_subprocess_list:
                        if sub_idx < len(st.session_state.get('processes', [])):
                            subprocess = st.session_state['processes'][sub_idx]
                            for stream_idx, stream in enumerate(subprocess.get('streams', [])):
                                # Only add if selected
                                stream_key = f"stream_{sub_idx}_{stream_idx}"
                                if selected_items.get(stream_key, True):
                                    all_streams.append(stream)
                    
                    if all_streams:
                        circle_y = y0 - 20
                        circle_spacing = 32
                        base_radius = 15
                        total_width = len(all_streams) * circle_spacing
                        start_x = int((x0 + x1) / 2 - total_width / 2 + circle_spacing / 2)
                        
                        for stream_idx, stream in enumerate(all_streams):
                            sv = stream.get('stream_values', {})
                            tin = sv.get('Tin', '') or stream.get('temp_in', '')
                            tout = sv.get('Tout', '') or stream.get('temp_out', '')
                            mdot = sv.get('ṁ', '') or stream.get('mdot', '')
                            cp_val = sv.get('cp', '') or stream.get('cp', '')
                            CP_val = sv.get('CP', '') or stream.get('CP', '')
                            
                            Q = 0
                            try:
                                if tin and tout:
                                    tin_f = float(tin)
                                    tout_f = float(tout)
                                    delta_T = abs(tout_f - tin_f)
                                    if CP_val:
                                        Q = float(CP_val) * delta_T
                                    elif mdot and cp_val:
                                        Q = float(mdot) * float(cp_val) * delta_T
                            except (ValueError, TypeError):
                                Q = 0
                            
                            if Q > 0:
                                import math as _m
                                # New size logic: below 1000 smaller, above 5000 very big, scaled in between
                                if Q < 1000:
                                    radius = base_radius + 5  # smaller
                                elif Q > 5000:
                                    radius = base_radius + 25  # very big
                                else:
                                    # Scale between 1000-5000
                                    scale_factor = (Q - 1000) / (5000 - 1000)  # 0 to 1
                                    radius = base_radius + 5 + int(scale_factor * 20)  # 5 to 25
                            else:
                                radius = base_radius
                            
                            # Determine color with temperature-based intensity
                            try:
                                if tin and tout:
                                    tin_f = float(tin)
                                    tout_f = float(tout)
                                    max_temp = max(tin_f, tout_f)
                                    
                                    if tin_f > tout_f:
                                        # Hot stream (Heat Source)
                                        if max_temp > 100:
                                            circle_color = (255, 0, 0, 220)  # Strong red
                                        else:
                                            circle_color = (255, 100, 100, 220)  # Less strong red
                                    else:
                                        # Cold Stream (Heat Sink)
                                        if max_temp > 100:
                                            circle_color = (0, 0, 255, 220)  # Strong blue
                                        else:
                                            circle_color = (100, 150, 255, 220)  # Less strong blue
                                else:
                                    circle_color = (150, 150, 150, 200)
                            except (ValueError, TypeError):
                                circle_color = (150, 150, 150, 200)
                            
                            circle_x = start_x + stream_idx * circle_spacing
                            base_img = draw_smooth_ellipse(
                                base_img,
                                [circle_x - radius, circle_y - radius, circle_x + radius, circle_y + radius],
                                fill=circle_color,
                                outline=(50, 50, 50, 255),
                                width=1,
                            )
                            draw = ImageDraw.Draw(base_img)
                    
                except (ValueError, TypeError):
                    continue
    
    return base_img

def generate_subprocess_level_map(group_idx=None):
    """Generate a map image showing subprocesses with all connections, energy data, colors, arrows.
    
    Args:
        group_idx: If provided, only show subprocesses from this process group. If None, show all.
    """
    import math
    
    snapshots_dict = st.session_state.get('map_snapshots', {})
    active_base = st.session_state.get('current_base', 'OpenStreetMap')
    
    if active_base == 'Blank':
        base_img = Image.new('RGBA', (800, 600), (242, 242, 243, 255))
    else:
        chosen_bytes = snapshots_dict.get(active_base) or st.session_state.get('map_snapshot')
        if not chosen_bytes:
            return None
        # Create a fresh copy of the base image to avoid modifying cached version
        original_img = Image.open(BytesIO(chosen_bytes)).convert("RGBA")
        base_img = original_img.copy()
    
    w, h = base_img.size
    draw = ImageDraw.Draw(base_img)
    
    # Load font - Cross-platform font loading
    BOX_FONT_SIZE = 20
    LABEL_FONT_SIZE = 6  # Very small font for stream labels

    # Try common font paths for different platforms
    font_paths = [
        # Linux common fonts
        "/usr/share/fonts/truetype/dejavu/DejaVuSans.ttf",
        "/usr/share/fonts/truetype/liberation/LiberationSans-Regular.ttf",
        "/usr/share/fonts/TTF/DejaVuSans.ttf",
        # macOS fonts
        "/System/Library/Fonts/Helvetica.ttc",
        "/System/Library/Fonts/Arial.ttf",
        # Windows fonts (fallback)
        "C:/Windows/Fonts/arial.ttf",
        # Generic fallback
        "DejaVuSans.ttf"
    ]

    font = None
    label_font = None

    for font_path in font_paths:
        try:
            font = ImageFont.truetype(font_path, BOX_FONT_SIZE)
            break  # Success, stop trying
        except (OSError, IOError):
            continue

    # Final fallback if all font paths fail
    if font is None:
        try:
            font = ImageFont.load_default()
        except:
            font = None

    # FORCE label_font to None to always use fallback sizing
    label_font = None

    # Create a tiny font for labels that will work everywhere
    try:
        # Try to create the smallest possible truetype font
        tiny_font = ImageFont.truetype("/usr/share/fonts/truetype/dejavu/DejaVuSans.ttf", 4)
    except:
        try:
            tiny_font = ImageFont.truetype("/System/Library/Fonts/Arial.ttf", 4)
        except:
            # Last resort: use default but we'll override the sizing
            tiny_font = None
    
    processes = st.session_state.get('processes', [])
    map_center = st.session_state.get('map_center', [0, 0])
    map_zoom = st.session_state.get('map_zoom', 17.5)
    proc_groups = st.session_state.get('proc_groups', [])
    proc_group_names = st.session_state.get('proc_group_names', [])
    
    if not processes:
        return None
    
    # Create mapping of subprocess index to group index
    subprocess_to_group = {}
    for g_idx, group_subprocess_list in enumerate(proc_groups):
        for subprocess_idx in group_subprocess_list:
            subprocess_to_group[subprocess_idx] = g_idx
    
    # Draw white canvas overlay (90% of screen, centered) - ALWAYS show for report
    overlay_w = int(w * 0.9)
    overlay_h = int(h * 0.9)
    center_px = w // 2
    center_py = h // 2
    overlay_x0 = int(center_px - overlay_w / 2)
    overlay_y0 = int(center_py - overlay_h / 2)
    overlay_x1 = overlay_x0 + overlay_w
    overlay_y1 = overlay_y0 + overlay_h
    
    margin = 20
    overlay_x0 = max(margin, overlay_x0)
    overlay_y0 = max(margin, overlay_y0)
    overlay_x1 = min(w - margin, overlay_x1)
    overlay_y1 = min(h - margin, overlay_y1)
    
    draw.rectangle([overlay_x0, overlay_y0, overlay_x1, overlay_y1], 
                 fill=(250, 250, 250, 40),
                 outline=(245, 245, 245, 80), 
                 width=1)
    
    # Add process area label
    if proc_group_names and group_idx is not None and group_idx < len(proc_group_names):
        overlay_label = f"Process Area: {proc_group_names[group_idx]}"
    elif proc_group_names:
        overlay_label = f"Process Area: {proc_group_names[0]}" if len(proc_group_names) == 1 else "Subprocess View"
    else:
        overlay_label = "Subprocess View"
    
    if font:
        label_bbox = draw.textbbox((0, 0), overlay_label, font=font)
        label_w = label_bbox[2] - label_bbox[0]
        label_h = label_bbox[3] - label_bbox[1]
    else:
        label_w = len(overlay_label) * 6
        label_h = 10
    
    label_x = overlay_x0 + 15
    label_y = overlay_y0 + 15
    
    draw.rectangle([label_x-5, label_y-3, label_x+label_w+5, label_y+label_h+3], 
                 fill=(255, 255, 255, 120), 
                 outline=(220, 220, 220, 100), 
                 width=1)
    
    if font:
        draw.text((label_x, label_y), overlay_label, fill=(40, 40, 40, 255), font=font)
    else:
        draw.text((label_x, label_y), overlay_label, fill=(40, 40, 40, 255))
    
    # Filter processes by group if specified
    if group_idx is not None:
        proc_groups_list = st.session_state.get('proc_groups', [])
        if group_idx < len(proc_groups_list):
            subprocess_indices = set(proc_groups_list[group_idx])  # Convert to set for fast lookup
            print(f"DEBUG: Generating subprocess map for group {group_idx}")
            print(f"  subprocess_indices (set): {subprocess_indices}")
            print(f"  proc_groups_list: {proc_groups_list}")
        else:
            subprocess_indices = set()
            print(f"DEBUG: Group {group_idx} is out of bounds, no subprocesses")
    else:
        subprocess_indices = set(range(len(processes)))
        print(f"DEBUG: Generating subprocess map for all processes: {subprocess_indices}")
    
    # Collect positioned subprocesses (filtered by group if specified)
    positioned = []
    name_index = {}
    skipped_count = 0
    included_count = 0
    
    print(f"  Starting subprocess iteration, total processes: {len(processes)}")
    
    for i, p in enumerate(processes):
        # Skip if filtering by group and this subprocess is not in the group
        if group_idx is not None and i not in subprocess_indices:
            print(f"    Process {i} ({p.get('name', 'unnamed')}): SKIPPED (not in group {group_idx})")
            skipped_count += 1
            continue
        
        print(f"    Process {i} ({p.get('name', 'unnamed')}): INCLUDED in group {group_idx}")
        included_count += 1
        lat = p.get('lat')
        lon = p.get('lon')
        if lat in (None, "", "None") or lon in (None, "", "None"):
            continue
        
        try:
            lat_f = float(lat)
            lon_f = float(lon)
            proc_px, proc_py = snapshot_lonlat_to_pixel(
                lon_f, lat_f,
                (map_center[1], map_center[0]),
                map_zoom, w, h
            )
            if proc_px < -50 or proc_py < -20 or proc_px > w + 50 or proc_py > h + 20:
                continue
            
            label = p.get('name') or f"P{i+1}"
            scale = float(p.get('box_scale', 6.0) or 6.0)
            base_padding = 18
            padding = int(base_padding * scale)
            text_bbox = draw.textbbox((0, 0), label, font=font) if font else (0, 0, len(label) * 6, 10)
            tw = text_bbox[2] - text_bbox[0]
            th = text_bbox[3] - text_bbox[1]
            box_w = int(tw * scale + padding * 2)
            box_h = int(th * scale + padding * 2)
            x0 = int(proc_px - box_w / 2)
            y0 = int(proc_py - box_h / 2)
            x1 = x0 + box_w
            y1 = y0 + box_h
            
            if x1 < 0 or y1 < 0 or x0 > w or y0 > h:
                continue
            
            positioned.append({
                'idx': i,
                'label': label,
                'center': (proc_px, proc_py),
                'box': (x0, y0, x1, y1),
                'next_raw': p.get('next', '') or '',
                'type': 'subprocess'
            })
            lname = label.strip().lower()
            name_index.setdefault(lname, []).append(len(positioned) - 1)
        except (ValueError, TypeError):
            continue
    
    # Helper: draw arrow with head
    def _draw_arrow(draw_ctx, x_start, y_start, x_end, y_end, color=(0, 0, 0, 255), width=3, head_len=18, head_angle_deg=30):
        draw_ctx.line([(x_start, y_start), (x_end, y_end)], fill=color, width=width, joint='curve')
        ang = math.atan2(y_end - y_start, x_end - x_start)
        ang_left = ang - math.radians(head_angle_deg)
        ang_right = ang + math.radians(head_angle_deg)
        x_left = x_end - head_len * math.cos(ang_left)
        y_left = y_end - head_len * math.sin(ang_left)
        x_right = x_end - head_len * math.cos(ang_right)
        y_right = y_end - head_len * math.sin(ang_right)
        draw_ctx.polygon([(x_end, y_end), (x_left, y_left), (x_right, y_right)], fill=color)
    
    def _resolve_targets(target_token, positioned_list, name_idx):
        target_token = target_token.strip()
        if not target_token:
            return []
        if target_token.isdigit():
            idx_int = int(target_token) - 1
            for d in positioned_list:
                if d['idx'] == idx_int:
                    return [d]
            return []
        lname2 = target_token.lower()
        if lname2 in name_idx:
            return [positioned_list[i] for i in name_idx[lname2]]
        return [d for d in positioned_list if d['label'].lower() == lname2]
    
    # Draw connection arrows and collect product stream info
    connection_product_streams = []
    
    for src in positioned:
        raw_next = src['next_raw']
        if not raw_next:
            continue
        parts = []
        for chunk in raw_next.replace(';', ',').replace('|', ',').split(','):
            part = chunk.strip()
            if part:
                parts.append(part)
        if not parts:
            continue
        
        sx, sy = src['center']
        for part_token in parts:
            targets = _resolve_targets(part_token, positioned, name_index)
            for tgt in targets:
                if tgt is src:
                    continue
                tx, ty = tgt['center']
                sx0, sy0, sx1, sy1 = src['box']
                sw2 = (sx1 - sx0) / 2.0
                sh2 = (sy1 - sy0) / 2.0
                tx0, ty0, tx1, ty1 = tgt['box']
                tw2 = (tx1 - tx0) / 2.0
                th2 = (ty1 - ty0) / 2.0
                vec_dx = tx - sx
                vec_dy = ty - sy
                if vec_dx == 0 and vec_dy == 0:
                    continue
                
                factors_s = []
                if vec_dx != 0:
                    factors_s.append(sw2 / abs(vec_dx))
                if vec_dy != 0:
                    factors_s.append(sh2 / abs(vec_dy))
                t_s = min(factors_s) if factors_s else 0
                
                factors_t = []
                if vec_dx != 0:
                    factors_t.append(tw2 / abs(vec_dx))
                if vec_dy != 0:
                    factors_t.append(th2 / abs(vec_dy))
                t_t = min(factors_t) if factors_t else 0
                
                start_x = sx + vec_dx * t_s * 1.02
                start_y = sy + vec_dy * t_s * 1.02
                end_x = tx - vec_dx * t_t * 1.02
                end_y = ty - vec_dy * t_t * 1.02
                _draw_arrow(draw, start_x, start_y, end_x, end_y, color=(0, 0, 0, 245), width=5)
                
                # Collect product streams for labeling
                if src.get('type') == 'subprocess':
                    src_proc = processes[src['idx']]
                    src_streams = src_proc.get('streams', []) or []
                    product_streams = [s for s in src_streams if s.get('type', 'product') == 'product']
                    if product_streams:
                        connection_product_streams.append({
                            'start_x': start_x, 'start_y': start_y,
                            'end_x': end_x, 'end_y': end_y,
                            'streams': product_streams
                        })
    
    # Draw product stream labels on connection arrows
    def _draw_connection_stream_label(draw_ctx, start_x, start_y, end_x, end_y, streams, font_obj):
        try:
            small_font = ImageFont.truetype("/System/Library/Fonts/Helvetica.ttc", 10)
        except:
            try:
                small_font = ImageFont.truetype("arial.ttf", 10)
            except:
                small_font = font_obj
        
        all_parts = []
        for s in streams:
            display_vars = s.get('display_vars', [])
            sv = s.get('stream_values', {}) or s.get('product_values', {})
            
            tin_val = sv.get('Tin', '') or s.get('temp_in', '')
            tout_val = sv.get('Tout', '') or s.get('temp_out', '')
            mdot_val = sv.get('ṁ', '') or s.get('mdot', '')
            cp_val = sv.get('cp', '') or s.get('cp', '')
            CP_val = sv.get('CP', '')
            
            # Calculate Q (heat power)
            Q_val = ''
            try:
                if tin_val and tout_val and mdot_val and cp_val:
                    Q_calc = float(mdot_val) * float(cp_val) * abs(float(tout_val) - float(tin_val))
                    Q_val = f"{Q_calc:.2f}"
            except (ValueError, TypeError):
                Q_val = ''
            
            stream_parts = []
            stream_name = s.get('name', '')
            if tin_val:
                stream_parts.append(f"Tin={tin_val}")
            if tout_val and "Tout" in display_vars:
                stream_parts.append(f"Tout={tout_val}")
            if mdot_val and "ṁ" in display_vars:
                stream_parts.append(f"ṁ={mdot_val}")
            if cp_val and "cp" in display_vars:
                stream_parts.append(f"cp={cp_val}")
            if CP_val and "CP" in display_vars:
                stream_parts.append(f"CP={CP_val}")
            if Q_val:
                stream_parts.append(f"Q={Q_val}kW")
            
            if stream_parts:
                label = " | ".join(stream_parts)
                if stream_name:
                    label = f"{stream_name}: {label}"
                all_parts.append(label)
        
        if not all_parts:
            return
        
        mid_x = (start_x + end_x) / 2
        mid_y = (start_y + end_y) / 2
        perp_offset = 18
        line_height = 14
        
        for line_idx, stream_label in enumerate(all_parts):
            if small_font:
                tb = draw_ctx.textbbox((0, 0), stream_label, font=small_font)
                t_width = tb[2] - tb[0]
                t_height = tb[3] - tb[1]
            else:
                t_width = len(stream_label) * 5
                t_height = 8
            
            lx = int(mid_x - t_width / 2)
            ly = int(mid_y + perp_offset + (line_idx * line_height))
            
            draw_ctx.rectangle([lx - 3, ly - 1, lx + t_width + 3, ly + t_height + 1], 
                               fill=(255, 255, 255, 235), outline=(120, 120, 120, 150))
            if small_font:
                draw_ctx.text((lx, ly), stream_label, fill=(0, 70, 0, 255), font=small_font)
            else:
                draw_ctx.text((lx, ly), stream_label, fill=(0, 70, 0, 255))
    
    # Draw subprocess boxes
    for item in positioned:
        x0, y0, x1, y1 = item['box']
        label = item['label']
        
        fill_color = (224, 242, 255, 245)  # Light blue
        border_color = (23, 105, 170, 255)  # Dark blue
        text_color = (10, 53, 85, 255)
        
        draw.rectangle([x0, y0, x1, y1], fill=fill_color, outline=border_color, width=2)
        
        box_w = x1 - x0
        box_h = y1 - y0
        if font:
            bbox_lbl = draw.textbbox((0, 0), label, font=font)
            t_w = bbox_lbl[2] - bbox_lbl[0]
            t_h = bbox_lbl[3] - bbox_lbl[1]
        else:
            t_w = len(label) * 6
            t_h = 10
        ct_x = int(x0 + (box_w - t_w) / 2)
        ct_y = int(y0 + (box_h - t_h) / 2)
        if font:
            draw.text((ct_x, ct_y), label, fill=text_color, font=font)
        else:
            draw.text((ct_x, ct_y), label, fill=text_color)
    
    # Stream labels removed - data shown in table below instead
    
    # Draw vertical stream arrows (Tin/Tout)
    arrow_len_v = 45
    base_stream_spacing = 35  # Even more reduced spacing
    
    def _draw_v_arrow(draw_ctx, x_pos, y_from, y_to, head_at_end=True, color=(0, 0, 0, 245), width=3):
        draw_ctx.line([(x_pos, y_from), (x_pos, y_to)], fill=color, width=width)
        head_len = 11
        head_half = 7
        if head_at_end:
            direction = 1 if y_to >= y_from else -1
            tip_y = y_to
            base_y = y_to - direction * head_len
        else:
            direction = 1 if y_from >= y_to else -1
            tip_y = y_from
            base_y = y_from - direction * head_len
        draw_ctx.polygon([
            (x_pos, tip_y),
            (x_pos - head_half, base_y),
            (x_pos + head_half, base_y)
        ], fill=color)
    
    def _label_centered(text_str, x_center, y_baseline, above=True):
        if not text_str:
            return
        # Try to use tiny font, otherwise use manual sizing
        if tiny_font:
            tb = draw.textbbox((0, 0), text_str, font=tiny_font)
            t_width = tb[2] - tb[0]
            t_height = tb[3] - tb[1]
            text_xc = int(x_center - t_width / 2)
            text_yc = int(y_baseline - (t_height if above else 0))
            draw.rectangle([text_xc - 1, text_yc - 1, text_xc + t_width + 1, text_yc + t_height + 1], fill=(255, 255, 255, 240))
            draw.text((text_xc, text_yc), text_str, fill=(0, 0, 0, 255), font=tiny_font)
        else:
            # Manual ultra-small sizing as fallback
            t_width = len(text_str) * 2.5
            t_height = 4
            text_xc = int(x_center - t_width / 2)
            text_yc = int(y_baseline - (t_height if above else 0))
            draw.rectangle([text_xc - 1, text_yc - 1, text_xc + t_width + 1, text_yc + t_height + 1], fill=(255, 255, 255, 240))
            draw.text((text_xc, text_yc), text_str, fill=(0, 0, 0, 255))
    
    # Draw stream arrows for each subprocess
    for item in positioned:
        if item.get('type') != 'subprocess':
            continue
        
        proc_idx = item['idx']
        proc = processes[proc_idx]
        streams = proc.get('streams', []) or []
        if not streams:
            continue
        
        non_product_streams = [s for s in streams if s.get('type', 'product') != 'product']
        if not non_product_streams:
            continue
        
        x0, y0, x1, y1 = item['box']
        box_center_x = (x0 + x1) / 2
        n_streams = len(non_product_streams)
        stream_h_spacing = max(28, min(base_stream_spacing, (x1 - x0 - 20) / max(1, n_streams))) if n_streams > 1 else 0
        
        for s_i, s in enumerate(non_product_streams):
            offset = (s_i - (n_streams - 1) / 2) * stream_h_spacing
            sx = int(box_center_x + offset)
            
            tin_raw = s.get('temp_in', '')
            tout_raw = s.get('temp_out', '')
            try:
                tin_val = float(str(tin_raw).strip())
            except (ValueError, TypeError):
                tin_val = None
            try:
                tout_val = float(str(tout_raw).strip())
            except (ValueError, TypeError):
                tout_val = None
            
            if tin_val is not None and tout_val is not None:
                is_cooling = tin_val > tout_val
                col = (200, 25, 25, 255) if is_cooling else (25, 80, 200, 255)
            else:
                col = (90, 90, 90, 255)
            
            inbound_bottom = y0 - 2
            inbound_top = inbound_bottom - arrow_len_v
            _draw_v_arrow(draw, sx, inbound_top, inbound_bottom, head_at_end=True, color=col, width=4)
            
            outbound_top = y1 + 2
            outbound_bottom = outbound_top + arrow_len_v
            _draw_v_arrow(draw, sx, outbound_top, outbound_bottom, head_at_end=True, color=col, width=4)
            
            mdot_raw = s.get('mdot', '')
            cp_raw = s.get('cp', '')
            m_symbol = "ṁ"
            mdot_part = f"{m_symbol}={mdot_raw}" if mdot_raw not in (None, '') else ''
            cp_part = f"cp={cp_raw}" if cp_raw not in (None, '') else ''
            
            tin_label = f"Tin={tin_raw}" if tin_raw not in ('', None) else 'Tin=?'
            tout_label = f"Tout={tout_raw}" if tout_raw not in ('', None) else 'Tout=?'
            
            top_components = [tin_label]
            if mdot_part:
                top_components.append(mdot_part)
            if cp_part:
                top_components.append(cp_part)
            bot_components = [tout_label]
            if mdot_part:
                bot_components.append(mdot_part)
            if cp_part:
                bot_components.append(cp_part)
            
            top_text = "  |  ".join(top_components)
            bot_text = "  |  ".join(bot_components)
        # Labels removed - only arrows shown in subprocess view
    print(f"  - Total processes in session: {len(processes)}")
    print(f"  - Skipped: {skipped_count}, Included: {included_count}")
    print(f"  - Positioned (drawable): {len(positioned)}")
    return base_img

def generate_report():
    """Generate an HTML report with process maps, notes, and pinch analysis results."""
    import base64
    import os
    import tempfile
    import csv
    
    # Load the logo SVG file
    logo_b64 = ""
    try:
        logo_path = os.path.join(os.path.dirname(os.path.dirname(os.path.dirname(__file__))), 'data', 'symbol.svg')
        with open(logo_path, 'r') as f:
            svg_content = f.read()
        logo_b64 = base64.b64encode(svg_content.encode('utf-8')).decode('utf-8')
    except Exception:
        pass
    
    # Generate process-level maps for both OpenStreetMap and Satellite
    # Temporarily save current base to restore later
    original_base = st.session_state.get('current_base', 'OpenStreetMap')
    
    process_map_osm_b64 = ""
    process_map_satellite_b64 = ""
    
    # Generate OpenStreetMap version
    st.session_state['current_base'] = 'OpenStreetMap'
    process_map_osm = generate_process_level_map()
    if process_map_osm:
        img_buffer = BytesIO()
        process_map_osm.save(img_buffer, format='PNG')
        img_buffer.seek(0)
        process_map_osm_b64 = base64.b64encode(img_buffer.getvalue()).decode('utf-8')
    
    # Generate Satellite version
    st.session_state['current_base'] = 'Satellite'
    process_map_satellite = generate_process_level_map()
    if process_map_satellite:
        img_buffer = BytesIO()
        process_map_satellite.save(img_buffer, format='PNG')
        img_buffer.seek(0)
        process_map_satellite_b64 = base64.b64encode(img_buffer.getvalue()).decode('utf-8')
    
    # Restore original base BEFORE generating subprocess maps
    st.session_state['current_base'] = original_base
    
    # Generate subprocess maps for each process group
    proc_groups = st.session_state.get('proc_groups', [])
    proc_group_names = st.session_state.get('proc_group_names', [])
    subprocess_maps_b64 = []
    
    print(f"\n=== STARTING SUBPROCESS MAP GENERATION ===")
    print(f"Total process groups: {len(proc_groups)}")
    print(f"Process groups structure: {proc_groups}")
    print(f"Process group names: {proc_group_names}")
    
    # CRITICAL CHECK: Verify groups contain different subprocesses
    all_subprocess_indices = []
    for g_idx, g_subprocs in enumerate(proc_groups):
        all_subprocess_indices.extend(g_subprocs)
        print(f"  Group {g_idx} ('{proc_group_names[g_idx] if g_idx < len(proc_group_names) else 'unnamed'}'): subprocesses {g_subprocs}")
    
    if len(all_subprocess_indices) != len(set(all_subprocess_indices)):
        print(f"WARNING: Some subprocesses appear in multiple groups! This will cause duplicate maps.")
        print(f"All subprocess indices: {all_subprocess_indices}")
        print(f"Unique subprocess indices: {set(all_subprocess_indices)}")
    
    for group_idx in range(len(proc_groups)):
        if len(proc_groups[group_idx]) > 0:  # Only generate if group has subprocesses
            print(f"\n--- Generating map for group {group_idx} ('{proc_group_names[group_idx] if group_idx < len(proc_group_names) else 'unnamed'}') ---")
            print(f"Subprocesses that SHOULD appear in this map: {proc_groups[group_idx]}")
            
            # Generate unique map for this specific group - force fresh generation
            subprocess_map = generate_subprocess_level_map(group_idx=group_idx)
            if subprocess_map:
                # Create a fresh buffer for each map
                img_buffer = BytesIO()
                subprocess_map.save(img_buffer, format='PNG')
                img_buffer.seek(0)
                img_data = img_buffer.getvalue()
                map_b64 = base64.b64encode(img_data).decode('utf-8')
                img_buffer.close()  # Ensure buffer is closed
                
                # Debug: Print first 50 chars of base64 to verify uniqueness
                print(f"Group {group_idx} map encoded: {len(map_b64)} chars, starts with: {map_b64[:50]}")
                
                group_name = proc_group_names[group_idx] if group_idx < len(proc_group_names) else f"Process {group_idx + 1}"
                subprocess_maps_b64.append({
                    'name': group_name, 
                    'image': map_b64, 
                    'group_idx': group_idx,
                    'subprocess_count': len(proc_groups[group_idx])
                })
                print(f"Added map for '{group_name}' to list (index {len(subprocess_maps_b64)-1})")
            else:
                print(f"WARNING: No map generated for group {group_idx}")
    
    print(f"\n=== SUBPROCESS MAP GENERATION COMPLETE ===")
    print(f"Total maps generated: {len(subprocess_maps_b64)}")
    for idx, m in enumerate(subprocess_maps_b64):
        print(f"  Map {idx}: {m['name']} (group {m['group_idx']}, {m['subprocess_count']} subprocesses)")
    
    # Legacy single subprocess map for backward compatibility
    subprocess_map_b64 = subprocess_maps_b64[0]['image'] if subprocess_maps_b64 else ""
    
    # Get notes
    project_notes = st.session_state.get('project_notes', '')
    notes_html = ""
    if project_notes:
        for para in project_notes.split('\n'):
            if para.strip():
                notes_html += f"<p>{para}</p>\n"
    else:
        notes_html = "<p><em>No notes recorded.</em></p>"
    
    # Build data collection table
    processes = st.session_state.get('processes', [])
    proc_group_names = st.session_state.get('proc_group_names', [])
    proc_groups = st.session_state.get('proc_groups', [])
    
    # Create mapping of subprocess to process
    subprocess_to_process = {}
    for group_idx, group_subprocess_list in enumerate(proc_groups):
        for subprocess_idx in group_subprocess_list:
            subprocess_to_process[subprocess_idx] = group_idx
    
    data_table_html = """
    <table class="data-table">
        <thead>
            <tr>
                <th>Process</th>
                <th>Subprocess</th>
                <th>Stream</th>
                <th>Type</th>
                <th>Tin (°C)</th>
                <th>Tout (°C)</th>
                <th>ṁ</th>
                <th>cp</th>
                <th>CP</th>
                <th>Q (kW)</th>
                <th>Water In</th>
                <th>Water Out</th>
                <th>Density</th>
                <th>Pressure</th>
            </tr>
        </thead>
        <tbody>
    """
    
    for subprocess_idx, subprocess in enumerate(processes):
        # Get process name
        process_idx = subprocess_to_process.get(subprocess_idx)
        if process_idx is not None and process_idx < len(proc_group_names):
            process_name = proc_group_names[process_idx]
        else:
            process_name = "Unknown"
        
        subprocess_name = subprocess.get('name', f'Subprocess {subprocess_idx + 1}')
        
        # Get streams
        streams = subprocess.get('streams', [])
        if streams:
            for stream_idx, stream in enumerate(streams):
                stream_name = stream.get('name', f'Stream {stream_idx + 1}')
                stream_type = stream.get('type', 'product')
                
                # Get stream values
                stream_values = stream.get('stream_values', {})
                tin = stream_values.get('Tin', '')
                tout = stream_values.get('Tout', '')
                mdot = stream_values.get('ṁ', '')
                cp_val = stream_values.get('cp', '')
                CP = stream_values.get('CP', '')
                water_in = stream_values.get('Water Content In', '')
                water_out = stream_values.get('Water Content Out', '')
                density = stream_values.get('Density', '')
                pressure = stream_values.get('Pressure', '')
                
                # Fallback to legacy fields if stream_values is empty
                if not stream_values:
                    tin = stream.get('temp_in', '')
                    tout = stream.get('temp_out', '')
                    mdot = stream.get('mdot', '')
                    cp_val = stream.get('cp', '')
                
                # Calculate Q (heat power) = ṁ × cp × ΔT
                Q = ''
                try:
                    if tin and tout and mdot and cp_val:
                        tin_f = float(tin)
                        tout_f = float(tout)
                        mdot_f = float(mdot)
                        cp_f = float(cp_val)
                        Q_calc = mdot_f * cp_f * abs(tout_f - tin_f)
                        Q = f"{Q_calc:.2f}"
                except (ValueError, TypeError):
                    Q = ''
                
                data_table_html += f"""
                <tr>
                    <td>{process_name}</td>
                    <td>{subprocess_name}</td>
                    <td>{stream_name}</td>
                    <td>{stream_type}</td>
                    <td>{tin}</td>
                    <td>{tout}</td>
                    <td>{mdot}</td>
                    <td>{cp_val}</td>
                    <td>{CP}</td>
                    <td>{Q}</td>
                    <td>{water_in}</td>
                    <td>{water_out}</td>
                    <td>{density}</td>
                    <td>{pressure}</td>
                </tr>
                """
        else:
            # Subprocess with no streams
            data_table_html += f"""
            <tr>
                <td>{process_name}</td>
                <td>{subprocess_name}</td>
                <td colspan="12"><em>No streams</em></td>
            </tr>
            """
    
    data_table_html += """
        </tbody>
    </table>
    """
    
    # Get pinch notes
    pinch_notes = st.session_state.get('pinch_notes', '')
    pinch_notes_html = ""
    if pinch_notes:
        for para in pinch_notes.split('\n'):
            if para.strip():
                pinch_notes_html += f"<p>{para}</p>\n"
    else:
        pinch_notes_html = "<p><em>No pinch analysis notes recorded.</em></p>"
    
    # =====================================================
    # PINCH ANALYSIS DATA FOR REPORT
    # =====================================================
    pinch_section_html = ""
    composite_plot_b64 = ""
    gcc_plot_b64 = ""
    interval_plot_b64 = ""
    
    if PINCH_AVAILABLE:
        try:
            # Helper function to extract stream info (same as main code)
            def get_stream_info_report(stream):
                properties = stream.get('properties', {})
                values = stream.get('values', {})
                stream_values = stream.get('stream_values', {})
                if not stream_values:
                    stream_values = stream.get('product_values', {})
                
                tin = None
                tout = None
                mdot = None
                cp_val = None
                CP_direct = None
                
                if stream_values:
                    if stream_values.get('Tin'):
                        try: tin = float(stream_values['Tin'])
                        except: pass
                    if stream_values.get('Tout'):
                        try: tout = float(stream_values['Tout'])
                        except: pass
                    if stream_values.get('ṁ'):
                        try: mdot = float(stream_values['ṁ'])
                        except: pass
                    if stream_values.get('cp'):
                        try: cp_val = float(stream_values['cp'])
                        except: pass
                    if stream_values.get('CP'):
                        try: CP_direct = float(stream_values['CP'])
                        except: pass
                
                if isinstance(properties, dict) and isinstance(values, dict):
                    for pk, pname in properties.items():
                        vk = pk.replace('prop', 'val')
                        v = values.get(vk, '')
                        if pname == 'Tin' and v and tin is None:
                            try: tin = float(v)
                            except: pass
                        elif pname == 'Tout' and v and tout is None:
                            try: tout = float(v)
                            except: pass
                        elif pname == 'ṁ' and v and mdot is None:
                            try: mdot = float(v)
                            except: pass
                        elif pname == 'cp' and v and cp_val is None:
                            try: cp_val = float(v)
                            except: pass
                        elif pname == 'CP' and v and CP_direct is None:
                            try: CP_direct = float(v)
                            except: pass
                
                if tin is None and stream.get('temp_in'):
                    try: tin = float(stream['temp_in'])
                    except: pass
                if tout is None and stream.get('temp_out'):
                    try: tout = float(stream['temp_out'])
                    except: pass
                if mdot is None and stream.get('mdot'):
                    try: mdot = float(stream['mdot'])
                    except: pass
                if cp_val is None and stream.get('cp'):
                    try: cp_val = float(stream['cp'])
                    except: pass
                
                stream_type = None
                if tin is not None and tout is not None:
                    stream_type = "Hot stream (Heat Source)" if tin > tout else "Cold stream (Heat sink)"
                
                CP_flow = CP_direct if CP_direct is not None else (mdot * cp_val if mdot and cp_val else None)
                Q = abs(CP_flow * (tout - tin)) if CP_flow and tin is not None and tout is not None else None
                
                return {'tin': tin, 'tout': tout, 'mdot': mdot, 'cp': cp_val, 'CP': CP_flow, 'Q': Q, 'type': stream_type}
            
            # Extract stream data from selections
            processes = st.session_state.get('processes', [])
            sel_items = st.session_state.get('selected_items', {})
            
            streams_data = []
            for sel_key, is_sel in sel_items.items():
                if not is_sel or not sel_key.startswith("stream_"):
                    continue
                parts_split = sel_key.split("_")
                p_idx, s_idx = int(parts_split[1]), int(parts_split[2])
                if p_idx < len(processes):
                    proc = processes[p_idx]
                    proc_streams = proc.get('streams', [])
                    if s_idx < len(proc_streams):
                        strm = proc_streams[s_idx]
                        info = get_stream_info_report(strm)
                        if info['tin'] is not None and info['tout'] is not None and info['CP'] is not None:
                            streams_data.append({
                                'name': f"{proc.get('name', f'Subprocess {p_idx + 1}')} - {strm.get('name', f'Stream {s_idx + 1}')}",
                                'CP': info['CP'], 'Tin': info['tin'], 'Tout': info['tout'], 'Q': info['Q'], 'type': info['type']
                            })
            
            if len(streams_data) >= 2:
                # Build stream list HTML for side-by-side layout
                stream_list_html = ""
                for sel_key, is_sel in sel_items.items():
                    if sel_key.startswith("stream_"):
                        parts_split = sel_key.split("_")
                        p_idx, s_idx = int(parts_split[1]), int(parts_split[2])
                        if p_idx < len(processes):
                            proc = processes[p_idx]
                            proc_streams = proc.get('streams', [])
                            if s_idx < len(proc_streams):
                                strm = proc_streams[s_idx]
                                info = get_stream_info_report(strm)
                                stream_name = f"{proc.get('name', f'Subprocess {p_idx + 1}')} - {strm.get('name', f'Stream {s_idx + 1}')}"
                                
                                # Determine if selected
                                selected_class = "selected" if is_sel else ""
                                checkbox_content = "✓" if is_sel else ""
                                
                                # Show Q if available
                                q_display = f"{info['Q']:.2f} kW" if info['Q'] is not None else "N/A"
                                type_display = "Hot stream (Heat Source)" if info['type'] == "Hot stream (Heat Source)" else ("Cold stream (Heat sink)" if info['type'] == "Cold stream (Heat sink)" else (info['type'] if info['type'] else "N/A"))
                                
                                stream_list_html += f"""
                                <div class="stream-item {selected_class}">
                                    <span class="stream-checkbox">{checkbox_content}</span>
                                    <div style="flex: 1;">
                                        <div><strong>{stream_name}</strong></div>
                                        <div style="font-size: 11px; color: #666;">
                                            Type: {type_display} | Q: {q_display}
                                        </div>
                                    </div>
                                </div>
                                """
                
                # Generate process map with circles for selected streams
                process_map_with_streams = generate_process_level_map()
                process_map_streams_b64 = ""
                if process_map_with_streams:
                    img_buffer = BytesIO()
                    process_map_with_streams.save(img_buffer, format='PNG')
                    img_buffer.seek(0)
                    process_map_streams_b64 = base64.b64encode(img_buffer.getvalue()).decode('utf-8')
                
                # Build streams table HTML
                streams_table_html = """
                <table class="streams-table">
                    <thead>
                        <tr><th>Stream</th><th>Tin (°C)</th><th>Tout (°C)</th><th>CP (kW/K)</th><th>Q (kW)</th><th>Type</th></tr>
                    </thead>
                    <tbody>
                """
                for s in streams_data:
                    type_badge = f'<span class="badge hot">Hot stream (Heat Source)</span>' if s['type'] == 'Hot stream (Heat Source)' else f'<span class="badge cold">Cold Stream (Heat Sink)</span>'
                    streams_table_html += f"""
                        <tr>
                            <td>{s['name']}</td>
                            <td>{s['Tin']:.1f}</td>
                            <td>{s['Tout']:.1f}</td>
                            <td>{s['CP']:.2f}</td>
                            <td>{s['Q']:.2f}</td>
                            <td>{type_badge}</td>
                        </tr>
                    """
                streams_table_html += "</tbody></table>"
                
                # Run pinch analysis
                tmin = st.session_state.get('tmin_input', 10.0)
                
                with tempfile.NamedTemporaryFile(mode='w', suffix='.csv', delete=False, newline='') as f:
                    writer = csv.writer(f)
                    writer.writerow(['Tmin', str(tmin)])
                    writer.writerow(['CP', 'TSUPPLY', 'TTARGET'])
                    for strm in streams_data:
                        writer.writerow([strm['CP'], strm['Tin'], strm['Tout']])
                    temp_csv_path = f.name
                
                try:
                    pinch_obj = Pinch(temp_csv_path, options={})
                    pinch_obj.shiftTemperatures()
                    pinch_obj.constructTemperatureInterval()
                    pinch_obj.constructProblemTable()
                    pinch_obj.constructHeatCascade()
                    pinch_obj.constructShiftedCompositeDiagram('EN')
                    pinch_obj.constructCompositeDiagram('EN')
                    pinch_obj.constructGrandCompositeCurve('EN')
                    
                    # Calculate Heat Recovery
                    total_hot_duty = sum(abs(s['Q']) for s in streams_data if s['Tin'] > s['Tout'])
                    heat_recovery = total_hot_duty - pinch_obj.hotUtility
                    
                    results = {
                        'hot_utility': pinch_obj.hotUtility,
                        'cold_utility': pinch_obj.coldUtility,
                        'pinch_temperature': pinch_obj.pinchTemperature,
                        'heat_recovery': heat_recovery,
                        'tmin': pinch_obj.tmin,
                        'composite_diagram': pinch_obj.compositeDiagram,
                        'grand_composite_curve': pinch_obj.grandCompositeCurve,
                        'heat_cascade': pinch_obj.heatCascade,
                        'temperatures': pinch_obj._temperatures,
                        'streams': list(pinch_obj.streams)
                    }
                    
                    # Generate Composite Curves plot using Plotly (interactive HTML)
                    diagram = results['composite_diagram']
                    fig1 = go.Figure()
                    fig1.add_trace(go.Scatter(x=diagram['hot']['H'], y=diagram['hot']['T'], mode='lines+markers', name='Hot streams (Heat sources)', line=dict(color='red', width=3), marker=dict(size=8)))
                    fig1.add_trace(go.Scatter(x=diagram['cold']['H'], y=diagram['cold']['T'], mode='lines+markers', name='Cold stream (Heat Sinks)', line=dict(color='blue', width=3), marker=dict(size=8)))
                    fig1.add_hline(y=results['pinch_temperature'], line_dash='dash', line_color='gray', annotation_text=f"Pinch: {results['pinch_temperature']:.1f}°C")
                    fig1.update_layout(title='Composite Curves', xaxis_title='Enthalpy H (kW)', yaxis_title='Temperature T (°C)', height=600, width=900, xaxis=dict(rangemode='tozero'), yaxis=dict(rangemode='tozero'))
                    composite_plot_html = fig1.to_html(full_html=False, include_plotlyjs='cdn')
                    
                    # Generate Grand Composite Curve plot using Plotly (interactive HTML)
                    gcc_H = results['grand_composite_curve']['H']
                    gcc_T = results['grand_composite_curve']['T']
                    heat_cascade = results['heat_cascade']
                    fig2 = go.Figure()
                    for i in range(len(gcc_H) - 1):
                        color = 'red' if i < len(heat_cascade) and heat_cascade[i]['deltaH'] > 0 else ('blue' if i < len(heat_cascade) and heat_cascade[i]['deltaH'] < 0 else 'gray')
                        fig2.add_trace(go.Scatter(x=[gcc_H[i], gcc_H[i+1]], y=[gcc_T[i], gcc_T[i+1]], mode='lines+markers', line=dict(color=color, width=3), marker=dict(size=8, color=color), showlegend=False))
                    fig2.add_hline(y=results['pinch_temperature'], line_dash='dash', line_color='gray', annotation_text=f"Pinch: {results['pinch_temperature']:.1f}°C")
                    fig2.add_vline(x=0, line_color='black', line_width=1, opacity=0.3)
                    fig2.update_layout(title='Grand Composite Curve', xaxis_title='Net ΔH (kW)', yaxis_title='Shifted Temperature (°C)', height=600, width=900, yaxis=dict(rangemode='tozero'))
                    gcc_plot_html = fig2.to_html(full_html=False, include_plotlyjs=False)
                    
                    # Generate Temperature Interval Diagram using Plotly (interactive HTML)
                    interval_plot_html = ""
                    temps = results['temperatures']
                    pinch_streams = results['streams']
                    if pinch_streams and temps:
                        fig3 = go.Figure()
                        num_streams = len(pinch_streams)
                        x_positions = [(i + 1) * 1.0 for i in range(num_streams)]
                        
                        # Draw horizontal temperature lines
                        for temperature in temps:
                            fig3.add_shape(type="line", x0=0, x1=num_streams + 1, y0=temperature, y1=temperature, line=dict(color="gray", width=1, dash="dot"))
                        
                        # Draw pinch line
                        fig3.add_shape(type="line", x0=0, x1=num_streams + 1, y0=results['pinch_temperature'], y1=results['pinch_temperature'], line=dict(color="black", width=2, dash="dash"))
                        
                        # Draw stream bars with arrows
                        for i, stream in enumerate(pinch_streams):
                            ss, st_temp = stream['ss'], stream['st']
                            color = 'red' if stream['type'] == 'HOT' else 'blue'
                            fig3.add_trace(go.Scatter(x=[x_positions[i], x_positions[i]], y=[ss, st_temp], mode='lines', line=dict(color=color, width=10), showlegend=False))
                            fig3.add_annotation(x=x_positions[i], y=st_temp, ax=x_positions[i], ay=ss, xref='x', yref='y', axref='x', ayref='y', showarrow=True, arrowhead=2, arrowsize=1.5, arrowwidth=3, arrowcolor=color)
                            fig3.add_annotation(x=x_positions[i], y=max(ss, st_temp) + (max(temps) - min(temps)) * 0.03, text=f"S{i+1}", showarrow=False, font=dict(size=12, color='white'), bgcolor=color)
                        
                        fig3.update_layout(title='Temperature Interval Diagram', xaxis=dict(title='Streams', showticklabels=False, range=[0, num_streams + 1]), yaxis=dict(title='Shifted Temperature (°C)'), height=600, width=900, showlegend=False)
                        interval_plot_html = fig3.to_html(full_html=False, include_plotlyjs=False)
                    
                    # Build pinch analysis section HTML
                    # =====================================================
                    # HEAT PUMP INTEGRATION ANALYSIS FOR REPORT
                    # =====================================================
                    hpi_section_html = ""
                    
                    try:
                        # Create temporary CSV for HPI analysis
                        with tempfile.NamedTemporaryFile(mode='w', suffix='.csv', delete=False, newline='') as f:
                            writer = csv.writer(f)
                            writer.writerow(['Tmin', str(tmin)])
                            writer.writerow(['CP', 'TSUPPLY', 'TTARGET'])
                            for strm in streams_data:
                                writer.writerow([strm['CP'], strm['Tin'], strm['Tout']])
                            hpi_csv_path = f.name
                        
                        try:
                            # Create Pinchmain instance
                            pyPinchHPI = Pinchmain(hpi_csv_path, options={})
                            
                            # Create HPI instance
                            hpi_analysis = HPI(hpi_csv_path, Tsinkout=None, pyPinch=pyPinchHPI)
                            
                            # Run HPI analysis
                            hpi_analysis.GCCdraw = pyPinchHPI.solvePinchforHPI().grandCompositeCurve
                            hpi_analysis.deleteTemperaturePockets()
                            hpi_analysis.GCCSource, hpi_analysis.GCCSink = hpi_analysis.splitHotandCold()
                            
                            # Check if HPI is possible
                            if hpi_analysis.GCCSource['T'] and hpi_analysis.GCCSink['T']:
                                # Run integration to get available heat pumps
                                hpi_analysis.IntegrateHeatPump()
                                hpi_analysis.findIntegration()
                                
                                # Get available heat pump types
                                all_hp_data = []
                                if hpi_analysis.IntegrationPoint and len(hpi_analysis.IntegrationPoint['Temp']) > 0:
                                    int_temp = hpi_analysis.IntegrationPoint['Temp'][-1]
                                    available_hps = hpi_analysis.get_available_heat_pumps(int_temp)
                                    
                                    total_hot_duty = sum(abs(s['Q']) for s in streams_data if s['Tin'] > s['Tout'])
                                    
                                    # Calculate integration for each available heat pump
                                    for hp in available_hps:
                                        if hp['available']:
                                            hpi_analysis.KoWP = []
                                            hpi_analysis.EvWP = []
                                            hpi_analysis.COPwerte = []
                                            hpi_analysis.COPT = []
                                            hpi_analysis.IntegrateHeatPump_specific(hp['name'])
                                            hpi_analysis.findIntegration()
                                            
                                            if hpi_analysis.IntegrationPoint and len(hpi_analysis.IntegrationPoint['Temp']) > 0:
                                                hp_int_temp = hpi_analysis.IntegrationPoint['Temp'][-1]
                                                hp_int_qsource = hpi_analysis.IntegrationPoint['QQuelle'][-1]
                                                hp_int_qsink = hpi_analysis.IntegrationPoint['QSenke'][-1]
                                                hp_int_cop = hpi_analysis.IntegrationPoint['COP'][-1]
                                                
                                                all_hp_data.append({
                                                    'name': hp['name'],
                                                    'cop': hp_int_cop,
                                                    't_source': hp_int_temp,
                                                    't_sink': hpi_analysis.Tsinkout,
                                                    'q_source': hp_int_qsource,
                                                    'q_sink': hp_int_qsink
                                                })
                                    
                                    # Sort by COP descending
                                    all_hp_data.sort(key=lambda x: x['cop'], reverse=True)
                                    
                                    if all_hp_data:
                                        # Build heat pump comparison table with dark-mode aware styling
                                        hp_table_html = """
                                        <style>
                                        /* Use inherited colors so table adapts to Streamlit theme */
                                        .hp-table{width:100%;border-collapse:collapse;font-size:12px;background-color:transparent;color:inherit;}
                                        .hp-table th{background-color:transparent;padding:8px;text-align:left;border:1px solid rgba(0,0,0,0.08);font-weight:bold;color:inherit;}
                                        .hp-table td{padding:8px;border:1px solid rgba(0,0,0,0.08);color:inherit;}
                                        .hp-table tr:nth-child(even){background-color:transparent;}
                                        @media (prefers-color-scheme: dark) {
                                            .hp-table th{border-color: rgba(255,255,255,0.06);}
                                            .hp-table td{border-color: rgba(255,255,255,0.06);}
                                        }
                                        </style>
                                        <table class="hp-table"><thead>
                                            <tr><th>Heat Pump</th><th>COP</th><th>T_source (°C)</th><th>T_sink (°C)</th><th>Q_source (kW)</th><th>Q_sink (kW)</th></tr>
                                        </thead>
                                        <tbody>
                                        """
                                        for hp in all_hp_data:
                                            hp_table_html += f"""
                                                <tr>
                                                    <td>{hp['name']}</td>
                                                    <td>{hp['cop']:.2f}</td>
                                                    <td>{hp['t_source']:.1f}</td>
                                                    <td>{hp['t_sink']:.1f}</td>
                                                    <td>{hp['q_source']:.1f}</td>
                                                    <td>{hp['q_sink']:.1f}</td>
                                                </tr>
                                            """
                                        hp_table_html += "</tbody></table>"
                                        
                                        # Generate HPI plot with all heat pumps and dropdown selector
                                        gcc_H = hpi_analysis.GCCdraw['H']
                                        gcc_T = hpi_analysis.GCCdraw['T']
                                        hp_colors = ['#00CC96', '#AB63FA', '#FFA15A', '#19D3F3', '#FF6692', '#B6E880', '#FF97FF', '#FECB52']
                                        
                                        # Create HPI figure
                                        fig_hpi = go.Figure()
                                        
                                        # Count GCC segments (they'll be visible for all heat pumps)
                                        num_gcc_segments = len(gcc_H) - 1
                                        
                                        # Plot GCC segments (always visible)
                                        for i in range(num_gcc_segments):
                                            if i < len(hpi_analysis.pyPinch.heatCascade):
                                                dh = hpi_analysis.pyPinch.heatCascade[i]['deltaH']
                                                color = 'red' if dh > 0 else ('blue' if dh < 0 else 'gray')
                                            else:
                                                color = 'gray'
                                            
                                            fig_hpi.add_trace(go.Scatter(
                                                x=[gcc_H[i], gcc_H[i+1]],
                                                y=[gcc_T[i], gcc_T[i+1]],
                                                mode='lines+markers',
                                                line=dict(color=color, width=2),
                                                marker=dict(size=5),
                                                showlegend=False,
                                                visible=True  # Always visible
                                            ))
                                        
                                        # Store integration data for all heat pumps
                                        hp_integration_traces = []
                                        
                                        # Add integration points for ALL available heat pumps
                                        for idx, hp_data in enumerate(all_hp_data):
                                            # Re-run integration for this heat pump
                                            hpi_analysis.KoWP = []
                                            hpi_analysis.EvWP = []
                                            hpi_analysis.COPwerte = []
                                            hpi_analysis.COPT = []
                                            hpi_analysis.IntegrateHeatPump_specific(hp_data['name'])
                                            hpi_analysis.findIntegration()
                                            
                                            if hpi_analysis.IntegrationPoint and len(hpi_analysis.IntegrationPoint['Temp']) > 0:
                                                color = hp_colors[idx % len(hp_colors)]
                                                hp_name = hp_data['name']
                                                int_temp = hpi_analysis.IntegrationPoint['Temp'][-1]
                                                int_qsource = hpi_analysis.IntegrationPoint['QQuelle'][-1]
                                                int_qsink = hpi_analysis.IntegrationPoint['QSenke'][-1]
                                                int_cop = hpi_analysis.IntegrationPoint['COP'][-1]
                                                t_sink = hpi_analysis.Tsinkout
                                                
                                                # Add source marker (visible only for first HP initially)
                                                fig_hpi.add_trace(go.Scatter(
                                                    x=[int_qsource], y=[int_temp],
                                                    mode='markers',
                                                    marker=dict(size=14, color=color, symbol='diamond', 
                                                              line=dict(width=2, color=color)),
                                                    name=f'{hp_name} - Source',
                                                    showlegend=True,
                                                    visible=(idx == 0),  # Only first HP visible by default
                                                    hovertemplate=f'<b>{hp_name} - Source</b><br>T: {int_temp:.1f}°C<br>Q: {int_qsource:.1f} kW<br>COP: {int_cop:.2f}<extra></extra>'
                                                ))
                                                
                                                # Add sink marker (visible only for first HP initially)
                                                fig_hpi.add_trace(go.Scatter(
                                                    x=[int_qsink], y=[t_sink],
                                                    mode='markers',
                                                    marker=dict(size=14, color=color, symbol='diamond-open',
                                                              line=dict(width=2, color=color)),
                                                    name=f'{hp_name} - Sink',
                                                    showlegend=True,
                                                    visible=(idx == 0),  # Only first HP visible by default
                                                    hovertemplate=f'<b>{hp_name} - Sink</b><br>T: {t_sink:.1f}°C<br>Q: {int_qsink:.1f} kW<br>COP: {int_cop:.2f}<extra></extra>'
                                                ))
                                                
                                                hp_integration_traces.append({
                                                    'name': hp_name,
                                                    'cop': int_cop,
                                                    'source_idx': len(fig_hpi.data) - 2,
                                                    'sink_idx': len(fig_hpi.data) - 1
                                                })
                                        
                                        # Create dropdown buttons for heat pump selection
                                        buttons = []
                                        for i, hp_trace in enumerate(hp_integration_traces):
                                            # Create visibility list: GCC always visible, only selected HP visible
                                            visible = [True] * num_gcc_segments  # GCC segments always visible
                                            for j in range(len(hp_integration_traces)):
                                                visible.append(j == i)  # source marker
                                                visible.append(j == i)  # sink marker
                                            
                                            buttons.append(dict(
                                                label=f"{hp_trace['name']} (COP: {hp_trace['cop']:.2f})",
                                                method="update",
                                                args=[{"visible": visible}]
                                            ))
                                        
                                        # Add pinch line
                                        fig_hpi.add_hline(y=pinch_obj.pinchTemperature, line_dash='dash', line_color='gray',
                                                        annotation_text=f"Pinch: {pinch_obj.pinchTemperature:.1f}°C")
                                        fig_hpi.add_vline(x=0, line_color='black', line_width=1, opacity=0.3)
                                        
                                        # Update layout with dropdown
                                        fig_hpi.update_layout(
                                            xaxis_title='Net ΔH (kW)',
                                            yaxis_title='Shifted Temperature (°C)',
                                            height=600,
                                            width=1100,
                                            xaxis=dict(domain=[0.35, 1], rangemode='tozero'),
                                            yaxis=dict(domain=[0, 1], rangemode='tozero'),
                                            showlegend=True,
                                            legend=dict(
                                                x=0.01,
                                                y=0.5,
                                                xanchor='left',
                                                yanchor='middle',
                                                bgcolor='rgba(255,255,255,0.8)',
                                                bordercolor='gray',
                                                borderwidth=1
                                            ),
                                            updatemenus=[dict(
                                                type="dropdown",
                                                x=0.01, xanchor="left",
                                                y=1, yanchor="top",
                                                buttons=buttons,
                                                bgcolor="white",
                                                bordercolor="gray",
                                                borderwidth=1
                                            )] if buttons else [],
                                            margin=dict(l=20, r=20, t=40, b=50)
                                        )
                                        
                                        hpi_plot_html = fig_hpi.to_html(full_html=False, include_plotlyjs=False)
                                        
                                        hpi_section_html = f"""
                                        <h3>Heat Pump Integration Analysis</h3>
                                        <div class="hpi-layout">
                                            <div class="hpi-table-section">
                                                <h4>🔍 Heat Pump Comparison</h4>
                                                {hp_table_html}
                                            </div>
                                            <div class="hpi-plot-section">
                                                {hpi_plot_html}
                                            </div>
                                        </div>
                                        """
                        finally:
                            os.unlink(hpi_csv_path)
                    except Exception as hpi_error:
                        hpi_section_html = f"<h3>Heat Pump Integration Analysis</h3><p><em>Heat pump integration could not be performed: {str(hpi_error)}</em></p>"
                    
                    pinch_section_html = f"""
                    <h2>📊 Potential Analysis</h2>
                    
                    <h3>Stream Selection & Process Map</h3>
                    <div class="stream-map-layout">
                        <div class="stream-list-section">
                            <h4>Selected Streams</h4>
                            {stream_list_html}
                        </div>
                        <div class="map-display-section">
                            <h4>Process Overview</h4>
                            {"<img src='data:image/png;base64," + process_map_streams_b64 + "' alt='Process Map with Streams' style='width: 80%; height: auto;'>" if process_map_streams_b64 else "<p>Map not available</p>"}
                        </div>
                    </div>
                    
                    <h3>Selected Streams (Detailed)</h3>
                    {streams_table_html}
                    
                    <h3>Pinch Analysis Results (ΔTmin = {tmin}°C)</h3>
                    <div class="metrics-row">
                        <div class="metric-card hot">
                            <div class="metric-label">Minimum Heating Demand</div>
                            <div class="metric-value">{results['hot_utility']:.2f} kW</div>
                        </div>
                        <div class="metric-card cold">
                            <div class="metric-label">Minimum Cooling Demand</div>
                            <div class="metric-value">{results['cold_utility']:.2f} kW</div>
                        </div>
                        <div class="metric-card pinch">
                            <div class="metric-label">Pinch Temperature</div>
                            <div class="metric-value">{results['pinch_temperature']:.1f} °C</div>
                        </div>
                        <div class="metric-card recovery">
                            <div class="metric-label">Heat Recovery Potential</div>
                            <div class="metric-value">{results['heat_recovery']:.2f} kW</div>
                        </div>
                    </div>
                    
                    <h3>Diagrams</h3>
                    <div class="plots-container">
                        <div class="plot-section">
                            {composite_plot_html}
                        </div>
                        <div class="plot-section">
                            {gcc_plot_html}
                        </div>
                    </div>
                    
                    {hpi_section_html}
                    
                    <h3>Temperature Interval Diagram</h3>
                    <div class="plots-container">
                        <div class="plot-section">
                            {interval_plot_html}
                        </div>
                    </div>
                    
                    <h3>📝 Pinch Analysis Notes</h3>
                    <div class="notes-section">
                        {pinch_notes_html}
                    </div>
                    """
                    
                finally:
                    os.unlink(temp_csv_path)
            else:
                pinch_section_html = """
                <h2>📊 Potential Analysis</h2>
                <p><em>Not enough streams selected for pinch analysis. Select at least 2 streams with complete data.</em></p>
                """
        except Exception as e:
            pinch_section_html = f"""
            <h2>📊 Potential Analysis</h2>
            <p><em>Error generating pinch analysis: {str(e)}</em></p>
            """
    else:
        pinch_section_html = """
        <h2>📊 Potential Analysis</h2>
        <p><em>Pinch analysis module not available.</em></p>
        """
    
    # Generate HTML with multi-page navigation
    html_content = f"""<!DOCTYPE html>
<html lang="en">
<head>
    <meta charset="UTF-8">
    <meta name="viewport" content="width=device-width, initial-scale=1.0">
    <title>Heat Integration Report</title>
    <style>
        body {{
            font-family: 'Segoe UI', Tahoma, Geneva, Verdana, sans-serif;
            max-width: 2600px;
            margin: 0 auto;
            padding: 20px;
            background-color: #f5f5f5;
        }}
        .report-container {{
            background-color: white;
            padding: 40px;
            border-radius: 8px;
            box-shadow: 0 2px 10px rgba(0,0,0,0.1);
        }}
        h1 {{
            color: #2c3e50;
            border-bottom: 3px solid #3498db;
            padding-bottom: 10px;
        }}
        h2 {{
            color: #34495e;
            margin-top: 40px;
            border-bottom: 2px solid #e0e0e0;
            padding-bottom: 8px;
        }}
        h3 {{
            color: #7f8c8d;
            margin-top: 25px;
        }}
        .timestamp {{
            color: #95a5a6;
            font-size: 14px;
            margin-bottom: 20px;
        }}
        
        /* Navigation Tabs */
        .nav-tabs {{
            display: flex;
            gap: 0;
            border-bottom: 3px solid #3498db;
            margin-bottom: 30px;
            padding: 0;
            list-style: none;
        }}
        .nav-tab {{
            padding: 15px 30px;
            background-color: #ecf0f1;
            cursor: pointer;
            border: none;
            font-size: 16px;
            font-weight: 600;
            color: #34495e;
            transition: all 0.3s ease;
            border-top-left-radius: 6px;
            border-top-right-radius: 6px;
            margin-bottom: -3px;
        }}
        .nav-tab:hover {{
            background-color: #d5dbdb;
        }}
        .nav-tab.active {{
            background-color: #3498db;
            color: white;
            border-bottom: 3px solid #3498db;
        }}
        
        /* Page Content */
        .page-content {{
            display: none;
        }}
        .page-content.active {{
            display: block;
            animation: fadeIn 0.3s ease-in;
        }}
        
        @keyframes fadeIn {{
            from {{ opacity: 0; }}
            to {{ opacity: 1; }}
        }}
        
        .maps-container {{
            display: flex;
            gap: 30px;
            justify-content: center;
            flex-wrap: nowrap;
            margin: 30px 0;
        }}
        .map-section {{
            text-align: center;
            flex: 1;
            min-width: 600px;
            max-width: 1100px;
        }}
        .map-section img {{
            max-width: 100%;
            width: 100%;
            border: 1px solid #ddd;
            border-radius: 4px;
            box-shadow: 0 2px 5px rgba(0,0,0,0.1);
        }}
        .map-section p {{
            font-weight: bold;
            color: #555;
            margin-top: 10px;
            font-size: 16px;
        }}
        .notes-section {{
            background-color: #fafafa;
            padding: 20px;
            border-radius: 4px;
            border-left: 4px solid #3498db;
            margin-top: 20px;
        }}
        .notes-section p {{
            margin: 8px 0;
            line-height: 1.6;
        }}
        .header-row {{
            display: flex;
            align-items: center;
            gap: 20px;
            margin-bottom: 10px;
        }}
        .header-logo {{
            height: 60px;
            width: auto;
        }}
        .streams-table {{
            width: 100%;
            border-collapse: collapse;
            margin: 20px 0;
        }}
        .streams-table th, .streams-table td {{
            border: 1px solid #ddd;
            padding: 12px;
            text-align: left;
        }}
        .streams-table th {{
            background-color: #f8f9fa;
            font-weight: bold;
        }}
        .streams-table tr:nth-child(even) {{
            background-color: #f9f9f9;
        }}
        .data-table {{
            width: 100%;
            border-collapse: collapse;
            margin: 20px 0;
            font-size: 11px;
        }}
        .data-table th, .data-table td {{
            border: 1px solid #ddd;
            padding: 8px;
            text-align: left;
        }}
        .data-table th {{
            background-color: #3498db;
            color: white;
            font-weight: bold;
        }}
        .data-table tr:nth-child(even) {{
            background-color: #f9f9f9;
        }}
        .data-table tr:hover {{
            background-color: #e3f2fd;
        }}
        .subprocess-maps-grid {{
            display: grid;
            grid-template-columns: repeat(2, 1fr);
            gap: 20px;
            margin: 30px 0;
        }}
        .subprocess-map-item {{
            text-align: center;
        }}
        .subprocess-map-item img {{
            max-width: 100%;
            width: 100%;
            border: 1px solid #ddd;
            border-radius: 4px;
            box-shadow: 0 2px 5px rgba(0,0,0,0.1);
        }}
        .subprocess-map-item p {{
            font-weight: bold;
            color: #555;
            margin-top: 10px;
            font-size: 14px;
        }}
        .badge {{
            padding: 4px 10px;
            border-radius: 4px;
            font-weight: bold;
            font-size: 12px;
        }}
        .badge.hot {{
            background-color: #ffebee;
            color: #c62828;
        }}
        .badge.cold {{
            background-color: #e3f2fd;
            color: #1565c0;
        }}
        .metrics-row {{
            display: flex;
            gap: 30px;
            margin: 30px 0;
            flex-wrap: wrap;
        }}
        .metric-card {{
            flex: 1;
            min-width: 200px;
            padding: 20px;
            border-radius: 8px;
            text-align: center;
        }}
        .metric-card.hot {{
            background-color: #ffebee;
            border: 2px solid #ef9a9a;
        }}
        .metric-card.cold {{
            background-color: #e3f2fd;
            border: 2px solid #90caf9;
        }}
        .metric-card.pinch {{
            background-color: #f3e5f5;
            border: 2px solid #ce93d8;
        }}
        .metric-card.recovery {{
            background-color: #e8f5e9;
            border: 2px solid #a5d6a7;
        }}
        .metric-label {{
            font-size: 14px;
            color: #666;
            margin-bottom: 8px;
        }}
        .metric-value {{
            font-size: 28px;
            font-weight: bold;
            color: #333;
        }}
        .plots-container {{
            display: flex;
            gap: 20px;
            justify-content: center;
            flex-wrap: nowrap;
            margin: 30px 0;
        }}
        .plot-section {{
            text-align: center;
            flex: 1;
            min-width: 0;
        }}
        .plot-section img {{
            max-width: 100%;
            border: 1px solid #ddd;
            border-radius: 4px;
            box-shadow: 0 2px 5px rgba(0,0,0,0.1);
        }}
        .hpi-layout {{
            display: flex;
            flex-direction: row;
            flex-wrap: nowrap;
            gap: 20px;
            margin: 30px 0;
            align-items: flex-start;
        }}
        .hpi-table-section {{
            flex: 0 0 35%;
            max-width: 35%;
            min-width: 300px;
        }}
        .hpi-plot-section {{
            flex: 1;
            min-width: 0;
            max-width: 65%;
        }}
        .hpi-plot-section > div {{
            width: 100% !important;
        }}
        /* Stream Selection and Map Layout */
        .stream-map-layout {{
            display: flex;
            gap: 20px;
            margin: 30px 0;
            align-items: flex-start;
        }}
        .stream-list-section {{
            flex: 0 0 50%;
            max-width: 50%;
        }}
        .map-display-section {{
            flex: 0 0 50%;
            max-width: 50%;
        }}
        .map-display-section img {{
            max-width: 100%;
            width: 100%;
            border: 1px solid #ddd;
            border-radius: 4px;
            box-shadow: 0 2px 5px rgba(0,0,0,0.1);
        }}
        .stream-item {{
            padding: 8px 12px;
            margin: 4px 0;
            border-radius: 4px;
            border: 1px solid #ddd;
            background-color: #f9f9f9;
            display: flex;
            align-items: center;
            gap: 10px;
        }}
        .stream-item.selected {{
            background-color: #e3f2fd;
            border-color: #2196F3;
        }}
        .stream-checkbox {{
            width: 16px;
            height: 16px;
            border: 2px solid #2196F3;
            border-radius: 3px;
            display: inline-block;
            text-align: center;
            line-height: 16px;
            font-size: 12px;
            color: #2196F3;
        }}
        @media print {{
            body {{
                background-color: white;
            }}
            .report-container {{
                box-shadow: none;
                padding: 20px;
            }}
            .nav-tabs {{
                display: none;
            }}
            .page-content {{
                display: block !important;
                page-break-before: always;
            }}
            .page-content:first-of-type {{
                page-break-before: auto;
            }}
        }}
    </style>
</head>
<body>
    <div class="report-container">
        <div class="header-row">
            <img src="data:image/svg+xml;base64,{logo_b64}" alt="Logo" class="header-logo">
            <h1 style="margin: 0; border: none; padding: 0;">Heat Integration Report</h1>
        </div>
        <p class="timestamp">Generated: {datetime.now().strftime('%Y-%m-%d %H:%M')}</p>
        
        <!-- Navigation Tabs -->
        <ul class="nav-tabs">
            <li class="nav-tab active" onclick="showPage('data-collection')">📍 Data Collection</li>
            <li class="nav-tab" onclick="showPage('potential-analysis')">📊 Potential Analysis</li>
        </ul>
        
        <!-- Page 1: Data Collection -->
        <div id="data-collection" class="page-content active">
            <h2>📍 Data Collection</h2>
            
            <h3>Process Level Maps</h3>
            <div class="maps-container">
                <div class="map-section">
                    {"<img src='data:image/png;base64," + process_map_osm_b64 + "' alt='Process - OpenStreetMap'>" if process_map_osm_b64 else "<p>Process map not available</p>"}
                    <p>OpenStreetMap</p>
                </div>
                <div class="map-section">
                    {"<img src='data:image/png;base64," + process_map_satellite_b64 + "' alt='Process - Satellite'>" if process_map_satellite_b64 else "<p>Satellite map not available</p>"}
                    <p>Satellite</p>
                </div>
            </div>
            
            <h3>Subprocess Level Maps</h3>
            {"<div class='subprocess-maps-grid'>" + "".join([f"<div class='subprocess-map-item'><img src='data:image/png;base64,{m['image']}' alt='{m['name']} Subprocess Map'><p>{m['name']}</p></div>" for m in subprocess_maps_b64]) + "</div>" if subprocess_maps_b64 else "<p>No subprocess maps available</p>"}
            
            <h3>📋 Collected Data</h3>
            {data_table_html}
            
            <h3>📝 Data Collection Notes</h3>
            <div class="notes-section">
                {notes_html}
            </div>
        </div>
        
        <!-- Page 2: Potential Analysis -->
        <div id="potential-analysis" class="page-content">
            {pinch_section_html}
        </div>
    </div>
    
    <script>
        function showPage(pageId) {{
            // Hide all pages
            const pages = document.querySelectorAll('.page-content');
            pages.forEach(page => page.classList.remove('active'));
            
            // Remove active class from all tabs
            const tabs = document.querySelectorAll('.nav-tab');
            tabs.forEach(tab => tab.classList.remove('active'));
            
            // Show selected page
            document.getElementById(pageId).classList.add('active');
            
            // Activate selected tab
            event.target.classList.add('active');
        }}
    </script>
</body>
</html>"""
    
    return html_content

# Title and Generate Report button
title_col, button_col = st.columns([4, 1])
with title_col:
    st.title("Potential Analysis")
with button_col:
    st.write("")  # Spacer to align button
    if st.button("📄 Generate Report", key="generate_report_btn"):
        try:
            html_data = generate_report()
            st.download_button(
                label="⬇️ Download Report",
                data=html_data,
                file_name=f"heat_integration_report_{datetime.now().strftime('%Y%m%d_%H%M')}.html",
                mime="text/html",
                key="download_report_btn"
            )
        except Exception as e:
            st.error(f"Error generating report: {str(e)}")

# Initialize session state for selections if not exists
if 'selected_items' not in st.session_state:
    st.session_state['selected_items'] = {}
if 'current_energy_demands' not in st.session_state:
    st.session_state['current_energy_demands'] = [
        {
            'heat_demand': 0.0,
            'cooling_demand': 0.0,
            'selected_streams': []  # Will be populated with all streams
        }
    ]

# Helper function to determine stream type and extract data
def get_stream_info(stream):
    """Extract Tin, Tout, mdot, cp, CP from stream and determine if Hot stream (Heat Source) or Cold Stream (Heat Sink).
    Calculate Q = CP * (Tout - Tin).
    CP can be provided directly, or calculated as mdot * cp.
    """
    properties = stream.get('properties', {})
    values = stream.get('values', {})
    
    # Also check stream_values (new structure)
    stream_values = stream.get('stream_values', {})
    if not stream_values:
        stream_values = stream.get('product_values', {})
    
    tin = None
    tout = None
    mdot = None
    cp_val = None
    CP_direct = None  # CP provided directly
    
    # First try stream_values (new structure)
    if stream_values:
        if 'Tin' in stream_values and stream_values['Tin']:
            try:
                tin = float(stream_values['Tin'])
            except (ValueError, TypeError):
                pass
        if 'Tout' in stream_values and stream_values['Tout']:
            try:
                tout = float(stream_values['Tout'])
            except (ValueError, TypeError):
                pass
        if 'ṁ' in stream_values and stream_values['ṁ']:
            try:
                mdot = float(stream_values['ṁ'])
            except (ValueError, TypeError):
                pass
        if 'cp' in stream_values and stream_values['cp']:
            try:
                cp_val = float(stream_values['cp'])
            except (ValueError, TypeError):
                pass
        if 'CP' in stream_values and stream_values['CP']:
            try:
                CP_direct = float(stream_values['CP'])
            except (ValueError, TypeError):
                pass
    
    # Check properties dict structure
    if isinstance(properties, dict) and isinstance(values, dict):
        for pk, pname in properties.items():
            vk = pk.replace('prop', 'val')
            v = values.get(vk, '')
            
            if pname == 'Tin' and v and tin is None:
                try:
                    tin = float(v)
                except (ValueError, TypeError):
                    pass
            elif pname == 'Tout' and v and tout is None:
                try:
                    tout = float(v)
                except (ValueError, TypeError):
                    pass
            elif pname == 'ṁ' and v and mdot is None:
                try:
                    mdot = float(v)
                except (ValueError, TypeError):
                    pass
            elif pname == 'cp' and v and cp_val is None:
                try:
                    cp_val = float(v)
                except (ValueError, TypeError):
                    pass
            elif pname == 'CP' and v and CP_direct is None:
                try:
                    CP_direct = float(v)
                except (ValueError, TypeError):
                    pass
    
    # Fallback to legacy fields
    if tin is None and stream.get('temp_in'):
        try:
            tin = float(stream['temp_in'])
        except (ValueError, TypeError):
            pass
    if tout is None and stream.get('temp_out'):
        try:
            tout = float(stream['temp_out'])
        except (ValueError, TypeError):
            pass
    if mdot is None and stream.get('mdot'):
        try:
            mdot = float(stream['mdot'])
        except (ValueError, TypeError):
            pass
    if cp_val is None and stream.get('cp'):
        try:
            cp_val = float(stream['cp'])
        except (ValueError, TypeError):
            pass
    
    # Determine stream type
    stream_type = None
    if tin is not None and tout is not None:
        if tin > tout:
            stream_type = "Hot stream (Heat Source)"
        else:
            stream_type = "Cold stream (Heat sink)"
    
    # Determine CP: use direct CP if provided, otherwise calculate from mdot * cp
    CP_flow = None
    if CP_direct is not None:
        CP_flow = CP_direct
    elif mdot is not None and cp_val is not None:
        CP_flow = mdot * cp_val
    
    # Calculate Q = CP * |Tout - Tin| (always positive)
    Q = None
    if CP_flow is not None and tin is not None and tout is not None:
        Q = abs(CP_flow * (tout - tin))
    
    return {
        'tin': tin,
        'tout': tout,
        'mdot': mdot,
        'cp': cp_val,
        'CP': CP_flow,
        'Q': Q,
        'type': stream_type
    }

# Get processes from session state
processes = st.session_state.get('processes', [])

if not processes:
    st.info("No processes found. Please add processes in the Data Collection page first.")
else:
    # Create side-by-side layout: streams on left, demands in middle, map on right
    streams_col, demands_col, map_col = st.columns([0.7, 0.8, 1.2])
    
    # Left column: Display streams with selection checkboxes
    with streams_col:
        st.markdown("**Streams Selection**")
        for idx, process in enumerate(processes):
            process_name = process.get('name', f'Subprocess {idx + 1}')
            
            # Only show process header if it has streams
            streams = process.get('streams', [])
            if streams:
                st.markdown(f"**{process_name}**")
                
                for stream_idx, stream in enumerate(streams):
                    stream_key = f"stream_{idx}_{stream_idx}"
                    if stream_key not in st.session_state['selected_items']:
                        st.session_state['selected_items'][stream_key] = True
                    
                    stream_cols_inner = st.columns([0.05, 0.25, 0.7])
                    stream_selected = stream_cols_inner[0].checkbox(
                        "S",
                        key=f"cb_{stream_key}",
                        value=st.session_state['selected_items'][stream_key],
                        label_visibility="collapsed"
                    )
                    st.session_state['selected_items'][stream_key] = stream_selected
                    
                    # Display stream name
                    stream_name = stream.get('name', f'Stream {stream_idx + 1}')
                    stream_cols_inner[1].write(stream_name)
                    
                    # Get stream info and display type + key values
                    info = get_stream_info(stream)
                    
                    display_parts = []
                    if info['tin'] is not None:
                        display_parts.append(f"Tin:{info['tin']}°C")
                    if info['tout'] is not None:
                        display_parts.append(f"Tout:{info['tout']}°C")
                    if info['CP'] is not None:
                        display_parts.append(f"CP:{info['CP']:.2f}")
                    if info['Q'] is not None:
                        display_parts.append(f"Q:{info['Q']:.2f} kW")
                    
                    if info['type']:
                        type_color = "🔴" if info['type'] == "Hot stream (Heat Source)" else "🔵"
                        display_parts.append(f"{type_color} {info['type']}")
                    
                    if display_parts:
                        stream_cols_inner[2].caption(' | '.join(display_parts))
                    else:
                        stream_cols_inner[2].caption("(incomplete data)")
    
    # Middle column: Current Energy Demands
    with demands_col:
        with st.expander("Current energy demand set", expanded=False):
            # Create stream options once
            stream_options = []
            for idx, process in enumerate(processes):
                for stream_idx, stream in enumerate(process.get('streams', [])):
                    stream_name = stream.get('name', f'Stream {stream_idx + 1} in {process.get('name', f'Subprocess {idx + 1}')}')
                    stream_options.append((idx, stream_idx, stream_name))
            
            all_stream_names = [name for _, _, name in stream_options]
            
            # Initialize selected_streams for the first demand if empty
            if not st.session_state['current_energy_demands'][0]['selected_streams']:
                st.session_state['current_energy_demands'][0]['selected_streams'] = all_stream_names.copy()
            
            # Display each demand
            demands_to_remove = []
            for i, demand in enumerate(st.session_state['current_energy_demands']):
                st.markdown(f"**Demand Set {i+1}**")
                
                col1, col2, col3 = st.columns([2, 2, 1])
                with col1:
                    demand['heat_demand'] = st.number_input(f"Heat Demand (kW) - Set {i+1}", min_value=0.0, step=0.1, value=demand['heat_demand'], key=f"heat_{i}")
                with col2:
                    demand['cooling_demand'] = st.number_input(f"Cooling Demand (kW) - Set {i+1}", min_value=0.0, step=0.1, value=demand['cooling_demand'], key=f"cooling_{i}")
                with col3:
                    if i > 0:  # Don't allow deleting the first demand
                        if st.button("🗑️", key=f"delete_{i}"):
                            demands_to_remove.append(i)
            
                demand['selected_streams'] = st.multiselect(
                    f"Streams - Set {i+1}", 
                    all_stream_names, 
                    default=demand['selected_streams'] if demand['selected_streams'] else all_stream_names,
                    key=f"streams_{i}"
                )
                st.markdown("---")
            
            # Remove demands marked for deletion
            for i in reversed(demands_to_remove):
                st.session_state['current_energy_demands'].pop(i)
            
            # Add new demand button
            if st.button("➕ Add Demand Set"):
                st.session_state['current_energy_demands'].append({
                    'heat_demand': 0.0,
                    'cooling_demand': 0.0,
                    'selected_streams': all_stream_names.copy()
                })
                st.rerun()
    
    # Right column: Display map with process circles
    with map_col:
        st.markdown("**🗺️ Georeferencing**")
        if st.session_state.get('map_snapshots'):
            map_img = generate_process_level_map()
            if map_img:
                st.image(map_img)
        else:
            st.info("Map not available. Lock a map in Data Collection first.")
    
    # Count selected streams
    selected_count = sum(1 for k, v in st.session_state['selected_items'].items() 
                         if v and k.startswith("stream_"))
    
    # =====================================================
    # PINCH ANALYSIS SECTION
    # =====================================================
    st.markdown("---")
    
    if not PINCH_AVAILABLE:
        st.error(f"Pinch analysis module not available: {PINCH_IMPORT_ERROR or 'Unknown error'}")
        st.info("Please ensure the pinch_tool module is properly installed.")
    else:
        # Helper function to extract stream data from selection
        def extract_stream_data(procs, sel_items):
            """
            Extract stream data from selected items.
            Returns list of dicts with: CP, Tin, Tout, Q
            CP can be provided directly, or calculated as mdot * cp.
            Q = CP * (Tout - Tin)
            """
            result_streams = []
            
            for sel_key, is_sel in sel_items.items():
                if not is_sel:
                    continue
                    
                if sel_key.startswith("stream_"):
                    parts_split = sel_key.split("_")
                    p_idx = int(parts_split[1])
                    s_idx = int(parts_split[2])
                    
                    if p_idx < len(procs):
                        proc = procs[p_idx]
                        proc_streams = proc.get('streams', [])
                        
                        if s_idx < len(proc_streams):
                            strm = proc_streams[s_idx]
                            
                            # Use get_stream_info to extract all values consistently
                            info = get_stream_info(strm)
                            
                            tin = info['tin']
                            tout = info['tout']
                            CP = info['CP']
                            Q = info['Q']
                            
                            # Only add if we have the required data
                            if tin is not None and tout is not None and CP is not None:
                                strm_name = strm.get('name', f'Stream {s_idx + 1}')
                                proc_nm = proc.get('name', f'Subprocess {p_idx + 1}')
                                result_streams.append({
                                    'name': f"{proc_nm} - {strm_name}",
                                    'CP': CP,
                                    'Tin': tin,
                                    'Tout': tout,
                                    'Q': Q
                                })
            
            return result_streams
        
        # Helper function to run pinch analysis
        def run_pinch_analysis(strm_data, delta_tmin):
            """
            Run pinch analysis on the given stream data.
            Returns the Pinch object with results.
            """
            # Create a temporary CSV file with the stream data
            with tempfile.NamedTemporaryFile(mode='w', suffix='.csv', delete=False, newline='') as f:
                writer = csv.writer(f)
                writer.writerow(['Tmin', str(delta_tmin)])
                writer.writerow(['CP', 'TSUPPLY', 'TTARGET'])
                
                for strm in strm_data:
                    writer.writerow([strm['CP'], strm['Tin'], strm['Tout']])
                
                temp_csv_path = f.name
            
            try:
                # Run pinch analysis without drawing (we'll draw ourselves)
                pinch_obj = Pinch(temp_csv_path, options={})
                pinch_obj.shiftTemperatures()
                pinch_obj.constructTemperatureInterval()
                pinch_obj.constructProblemTable()
                pinch_obj.constructHeatCascade()
                pinch_obj.constructShiftedCompositeDiagram('EN')
                pinch_obj.constructCompositeDiagram('EN')
                pinch_obj.constructGrandCompositeCurve('EN')
                
                return pinch_obj
            finally:
                # Clean up temp file
                os.unlink(temp_csv_path)
        
        # Extract stream data from selections
        streams_data = extract_stream_data(processes, st.session_state['selected_items'])
        
        if len(streams_data) < 2:
            st.info("Select at least 2 streams with complete data (Tin, Tout, and either CP or ṁ+cp) to run pinch analysis.")
            
            # Show what data is missing for selected streams
            if selected_count > 0:
                st.markdown("**Data status for selected items:**")
                for sel_key, is_sel in st.session_state['selected_items'].items():
                    if not is_sel:
                        continue
                    if sel_key.startswith("stream_"):
                        parts_split = sel_key.split("_")
                        p_idx = int(parts_split[1])
                        s_idx = int(parts_split[2])
                        
                        if p_idx < len(processes):
                            proc = processes[p_idx]
                            proc_streams = proc.get('streams', [])
                            
                            if s_idx < len(proc_streams):
                                strm = proc_streams[s_idx]
                                strm_name = strm.get('name', f'Stream {s_idx + 1}')
                                proc_nm = proc.get('name', f'Subprocess {p_idx + 1}')
                                
                                # Check what data is available
                                props = strm.get('properties', {})
                                vals = strm.get('values', {})
                                
                                has_tin = False
                                has_tout = False
                                has_mdot = False
                                has_cp = False
                                has_CP = False  # CP = ṁ * cp (heat capacity rate)
                                
                                # Check stream_values (new structure)
                                stream_vals = strm.get('stream_values', {})
                                if not stream_vals:
                                    stream_vals = strm.get('product_values', {})
                                
                                if stream_vals:
                                    if stream_vals.get('Tin'):
                                        has_tin = True
                                    if stream_vals.get('Tout'):
                                        has_tout = True
                                    if stream_vals.get('ṁ'):
                                        has_mdot = True
                                    if stream_vals.get('cp'):
                                        has_cp = True
                                    if stream_vals.get('CP'):
                                        has_CP = True
                                
                                if isinstance(props, dict) and isinstance(vals, dict):
                                    for pk, pname in props.items():
                                        vk = pk.replace('prop', 'val')
                                        v = vals.get(vk, '')
                                        if pname == 'Tin' and v:
                                            has_tin = True
                                        elif pname == 'Tout' and v:
                                            has_tout = True
                                        elif pname == 'ṁ' and v:
                                            has_mdot = True
                                        elif pname == 'cp' and v:
                                            has_cp = True
                                        elif pname == 'CP' and v:
                                            has_CP = True
                                
                                # Fallback to legacy
                                if not has_tin and strm.get('temp_in'):
                                    has_tin = True
                                if not has_tout and strm.get('temp_out'):
                                    has_tout = True
                                if not has_mdot and strm.get('mdot'):
                                    has_mdot = True
                                if not has_cp and strm.get('cp'):
                                    has_cp = True
                                
                                missing = []
                                if not has_tin:
                                    missing.append("Tin")
                                if not has_tout:
                                    missing.append("Tout")
                                # Either CP is provided directly, or both ṁ and cp are needed
                                if not has_CP and not (has_mdot and has_cp):
                                    if not has_mdot:
                                        missing.append("ṁ")
                                    if not has_cp:
                                        missing.append("cp")
                                    if not missing or (not has_mdot and not has_cp):
                                        # If neither ṁ nor cp, suggest CP as alternative
                                        missing.append("(or CP)")
                                
                                if missing:
                                    st.warning(f"⚠️ {proc_nm} - {strm_name}: Missing {', '.join(missing)}")
                                else:
                                    st.success(f"✅ {proc_nm} - {strm_name}: Complete data")
        else:
            # Auto-run pinch analysis
            try:
                # Row: Shifted toggle | ΔTmin (small) | spacer | Hot Utility | Cold Utility | Pinch Temp | Heat Recovery
                toggle_col, tmin_col, spacer, metric1, metric2, metric3, metric4 = st.columns([0.6, 0.5, 0.3, 0.6, 0.6, 0.6, 0.7])
                
                with toggle_col:
                    show_shifted = st.toggle("Show Shifted Composite Curves", value=False, key="shifted_toggle")
                
                with tmin_col:
                    tmin = st.number_input(
                        "ΔTmin",
                        min_value=1.0,
                        max_value=50.0,
                        value=10.0,
                        step=1.0,
                        key="tmin_input",
                        format="%.0f"
                    )
                
                pinch = run_pinch_analysis(streams_data, tmin)
                
                # Calculate Heat Recovery: Total heat available from hot streams minus hot utility needed
                total_hot_duty = sum(abs(s['Q']) for s in streams_data if s['Tin'] > s['Tout'])
                heat_recovery = total_hot_duty - pinch.hotUtility
                
                results = {
                    'hot_utility': pinch.hotUtility,
                    'cold_utility': pinch.coldUtility,
                    'pinch_temperature': pinch.pinchTemperature,
                    'heat_recovery': heat_recovery,
                    'tmin': pinch.tmin,
                    'composite_diagram': pinch.compositeDiagram,
                    'shifted_composite_diagram': pinch.shiftedCompositeDiagram,
                    'grand_composite_curve': pinch.grandCompositeCurve,
                    'heat_cascade': pinch.heatCascade,
                    'unfeasible_heat_cascade': pinch.unfeasibleHeatCascade,
                    'problem_table': pinch.problemTable,
                    'temperatures': pinch._temperatures,
                    'streams': list(pinch.streams)
                }
                
                metric1.metric("Minimum Heating Demand", f"{results['hot_utility']:.2f} kW")
                metric2.metric("Minimum Cooling Demand", f"{results['cold_utility']:.2f} kW")
                metric3.metric("Pinch Temperature", f"{results['pinch_temperature']:.1f} °C")
                metric4.metric("Heat Recovery Potential", f"{results['heat_recovery']:.2f} kW")
                
                # Side by side plots: Composite Curves (left) and Grand Composite Curve (right)
                plot_col1, plot_col2 = st.columns(2)
                
                # Build hover text for streams
                hot_streams = [s for s in streams_data if s['Tin'] > s['Tout']]
                cold_streams = [s for s in streams_data if s['Tin'] < s['Tout']]
                
                with plot_col1:
                    fig1 = go.Figure()
                    
                    # Select which diagram to show
                    if show_shifted:
                        diagram = results['shifted_composite_diagram']
                        curve_label = "Shifted"
                        title_text = "Shifted Composite Curves"
                        # For shifted, temperatures are shifted by ±Tmin/2
                        tmin_half = results['tmin'] / 2
                    else:
                        diagram = results['composite_diagram']
                        curve_label = ""
                        title_text = "Composite Curves"
                        tmin_half = 0
                    
                    # Hot stream (Heat Source) composite curve with hover info
                    hot_T = diagram['hot']['T']
                    hot_H = diagram['hot']['H']
                    
                    # Create hover text for hot stream (Heat Source) curve points
                    hot_hover = []
                    for i, (h, t) in enumerate(zip(hot_H, hot_T)):
                        # Find streams at this temperature (adjust for shifted temps)
                        if show_shifted:
                            actual_t = t + tmin_half  # Convert back to actual temp
                        else:
                            actual_t = t
                        matching = [s['name'] for s in hot_streams if min(s['Tin'], s['Tout']) <= actual_t <= max(s['Tin'], s['Tout'])]
                        stream_info = '<br>'.join(matching) if matching else 'Composite'
                        label = f"<b>Hot stream (Heat Source) {curve_label}</b>" if curve_label else "<b>Hot stream (Heat Source) Composite</b>"
                        hot_hover.append(f"{label}<br>T: {t:.1f}°C<br>H: {h:.1f} kW<br>Streams: {stream_info}")
                    
                    fig1.add_trace(go.Scatter(
                        x=hot_H, y=hot_T,
                        mode='lines+markers',
                        name='Hot streams (Heat sources)',
                        line=dict(color='red', width=2),
                        marker=dict(size=6),
                        hovertemplate='%{text}<extra></extra>',
                        text=hot_hover
                    ))
                    
                    # Cold Stream (Heat Sink) composite curve with hover info
                    cold_T = diagram['cold']['T']
                    cold_H = diagram['cold']['H']
                    
                    # Create hover text for Cold Stream (Heat Sink) curve points
                    cold_hover = []
                    for i, (h, t) in enumerate(zip(cold_H, cold_T)):
                        if show_shifted:
                            actual_t = t - tmin_half  # Convert back to actual temp
                        else:
                            actual_t = t
                        matching = [s['name'] for s in cold_streams if min(s['Tin'], s['Tout']) <= actual_t <= max(s['Tin'], s['Tout'])]
                        stream_info = '<br>'.join(matching) if matching else 'Composite'
                        label = f"<b>Cold streams (Heat sinks) {curve_label}</b>" if curve_label else "<b>Cold streams (Heat sinks) Composite</b>"
                        cold_hover.append(f"{label}<br>T: {t:.1f}°C<br>H: {h:.1f} kW<br>Streams: {stream_info}")
                    
                    fig1.add_trace(go.Scatter(
                        x=cold_H, y=cold_T,
                        mode='lines+markers',
                        name='Cold streams (Heat sinks)',
                        line=dict(color='blue', width=2),
                        marker=dict(size=6),
                        hovertemplate='%{text}<extra></extra>',
                        text=cold_hover
                    ))
                    
                    # Pinch temperature line
                    fig1.add_hline(
                        y=results['pinch_temperature'],
                        line_dash='dash',
                        line_color='gray',
                        annotation_text=f"Pinch: {results['pinch_temperature']:.1f}°C",
                        annotation_position='top right'
                    )
                    
                    fig1.update_layout(
                        title=dict(text=title_text, font=dict(size=14)),
                        xaxis_title='Enthalpy H (kW)',
                        yaxis_title='Temperature T (°C)',
                        height=400,
                        margin=dict(l=60, r=20, t=40, b=50),
                        legend=dict(x=0.7, y=0.1),
                        hovermode='closest',
                        xaxis=dict(rangemode='tozero'),
                        yaxis=dict(rangemode='tozero')
                    )
                    
                    st.plotly_chart(fig1, width='stretch', key="composite_chart")
                
                with plot_col2:
                    fig2 = go.Figure()
                    
                    gcc_H = results['grand_composite_curve']['H']
                    gcc_T = results['grand_composite_curve']['T']
                    heat_cascade = results['heat_cascade']
                    temperatures = results['temperatures']
                    
                    # Create hover text for GCC points
                    gcc_hover = []
                    for i, (h, t) in enumerate(zip(gcc_H, gcc_T)):
                        if i < len(heat_cascade):
                            dh = heat_cascade[i]['deltaH']
                            region = 'Heat deficit (needs heating)' if dh > 0 else ('Heat surplus (needs cooling)' if dh < 0 else 'Balanced')
                        else:
                            region = ''
                        gcc_hover.append(f"<b>GCC</b><br>Shifted T: {t:.1f}°C<br>Net ΔH: {h:.1f} kW<br>{region}")
                    
                    # Plot GCC with color segments
                    for i in range(len(gcc_H) - 1):
                        if i < len(heat_cascade):
                            if heat_cascade[i]['deltaH'] > 0:
                                color = 'red'
                            elif heat_cascade[i]['deltaH'] < 0:
                                color = 'blue'
                            else:
                                color = 'gray'
                        else:
                            color = 'gray'
                        
                        fig2.add_trace(go.Scatter(
                            x=[gcc_H[i], gcc_H[i+1]],
                            y=[gcc_T[i], gcc_T[i+1]],
                            mode='lines+markers',
                            line=dict(color=color, width=2),
                            marker=dict(size=6, color=color),
                            hovertemplate='%{text}<extra></extra>',
                            text=[gcc_hover[i], gcc_hover[i+1] if i+1 < len(gcc_hover) else ''],
                            showlegend=False
                        ))
                    
                    # Pinch temperature line
                    fig2.add_hline(
                        y=results['pinch_temperature'],
                        line_dash='dash',
                        line_color='gray',
                        annotation_text=f"Pinch: {results['pinch_temperature']:.1f}°C",
                        annotation_position='top right'
                    )
                    
                    # Zero enthalpy line
                    fig2.add_vline(x=0, line_color='black', line_width=1, opacity=0.3)
                    
                    fig2.update_layout(
                        title=dict(text='Grand Composite Curve', font=dict(size=14)),
                        xaxis_title='Net ΔH (kW)',
                        yaxis_title='Shifted Temperature (°C)',
                        height=400,
                        margin=dict(l=60, r=20, t=40, b=50),
                        hovermode='closest',
                        yaxis=dict(rangemode='tozero')
                    )
                    
                    st.plotly_chart(fig2, width='stretch', key="gcc_chart")
                
                # Heat Pump Integration Analysis
                st.markdown("---")
                st.markdown("#### Heat Pump Integration Analysis")
                
                try:
                    # Create temporary CSV for HPI analysis (same as pinch analysis)
                    with tempfile.NamedTemporaryFile(mode='w', suffix='.csv', delete=False, newline='') as f:
                        writer = csv.writer(f)
                        writer.writerow(['Tmin', str(tmin)])
                        writer.writerow(['CP', 'TSUPPLY', 'TTARGET'])
                        for strm in streams_data:
                            writer.writerow([strm['CP'], strm['Tin'], strm['Tout']])
                        hpi_csv_path = f.name
                    
                    try:
                        # Create Pinchmain instance
                        pyPinchHPI = Pinchmain(hpi_csv_path, options={})
                        
                        # Create HPI instance (Tsinkout=None for iterative analysis)
                        hpi_analysis = HPI(hpi_csv_path, Tsinkout=None, pyPinch=pyPinchHPI)
                        
                        # Run HPI analysis to get data (without plotting)
                        hpi_analysis.GCCdraw = pyPinchHPI.solvePinchforHPI().grandCompositeCurve
                        hpi_analysis.deleteTemperaturePockets()
                        hpi_analysis.GCCSource, hpi_analysis.GCCSink = hpi_analysis.splitHotandCold()
                        
                        # Check if there are sufficient hot and cold streams for heat pump integration
                        if not hpi_analysis.GCCSource['T'] or not hpi_analysis.GCCSink['T']:
                            st.warning("⚠️ Heat pump integration is not possible with the selected streams. The Grand Composite Curve does not have both heating and cooling requirements suitable for heat pump integration.")
                            raise ValueError("Insufficient streams for heat pump integration")
                        
                        # First run to get available heat pumps
                        hpi_analysis.IntegrateHeatPump()
                        hpi_analysis.findIntegration()
                        
                        # Get available heat pump types at the integration point
                        all_hp_data = []
                        if hpi_analysis.IntegrationPoint and len(hpi_analysis.IntegrationPoint['Temp']) > 0:
                            int_temp = hpi_analysis.IntegrationPoint['Temp'][-1]
                            available_hps = hpi_analysis.get_available_heat_pumps(int_temp)
                            
                            # Calculate total heat available from hot streams
                            total_hot_duty = sum(abs(s['Q']) for s in streams_data if s['Tin'] > s['Tout'])
                            
                            # Calculate integration for each available heat pump
                            for hp in available_hps:
                                if hp['available']:
                                    # Run integration for this specific heat pump
                                    hpi_analysis.KoWP = []
                                    hpi_analysis.EvWP = []
                                    hpi_analysis.COPwerte = []
                                    hpi_analysis.COPT = []
                                    hpi_analysis.IntegrateHeatPump_specific(hp['name'])
                                    hpi_analysis.findIntegration()
                                    
                                    # Get integration results
                                    if hpi_analysis.IntegrationPoint and len(hpi_analysis.IntegrationPoint['Temp']) > 0:
                                        hp_int_temp = hpi_analysis.IntegrationPoint['Temp'][-1]
                                        hp_int_qsource = hpi_analysis.IntegrationPoint['QQuelle'][-1]
                                        hp_int_qsink = hpi_analysis.IntegrationPoint['QSenke'][-1]
                                        hp_int_cop = hpi_analysis.IntegrationPoint['COP'][-1]
                                        coverage = (hp_int_qsource / total_hot_duty * 100) if total_hot_duty > 0 else 0
                                        
                                        all_hp_data.append({
                                            'name': hp['name'],
                                            'cop': hp_int_cop,
                                            't_source': hp_int_temp,
                                            't_sink': hpi_analysis.Tsinkout,
                                            'q_source': hp_int_qsource,
                                            'q_sink': hp_int_qsink,
                                            'coverage': coverage,
                                            'available': True
                                        })
                                else:
                                    # Unavailable heat pump
                                    all_hp_data.append({
                                        'name': hp['name'],
                                        'cop': None,
                                        't_source': None,
                                        't_sink': None,
                                        'q_source': None,
                                        'q_sink': None,
                                        'coverage': None,
                                        'available': False,
                                        'reason': hp['reason']
                                    })
                            
                            # Sort by COP descending
                            available_hps_sorted = sorted(
                                [hp for hp in all_hp_data if hp['available']], 
                                key=lambda x: x['cop'], 
                                reverse=True
                            )
                            available_hp_names = [hp['name'] for hp in available_hps_sorted]
                            
                            # Store integration data for all available heat pumps initially
                            all_hp_integration_data = []
                            for hp_name in available_hp_names:
                                # Re-run integration with this specific heat pump type
                                hpi_analysis.KoWP = []
                                hpi_analysis.EvWP = []
                                hpi_analysis.COPwerte = []
                                hpi_analysis.COPT = []
                                hpi_analysis.IntegrateHeatPump_specific(hp_name)
                                hpi_analysis.findIntegration()
                                
                                # Store the integration data
                                if hpi_analysis.IntegrationPoint and len(hpi_analysis.IntegrationPoint['Temp']) > 0:
                                    all_hp_integration_data.append({
                                        'name': hp_name,
                                        'int_temp': hpi_analysis.IntegrationPoint['Temp'][-1],
                                        'int_qsource': hpi_analysis.IntegrationPoint['QQuelle'][-1],
                                        'int_qsink': hpi_analysis.IntegrationPoint['QSenke'][-1],
                                        'int_cop': hpi_analysis.IntegrationPoint['COP'][-1],
                                        't_sink': hpi_analysis.Tsinkout
                                    })
                        else:
                            available_hp_names = []
                            all_hp_integration_data = []
                        
                        # Get HPI data
                        gcc_H = hpi_analysis.GCCdraw['H']
                        gcc_T = hpi_analysis.GCCdraw['T']
                        integration_point = hpi_analysis.IntegrationPoint
                        
                        # Define colors for different heat pumps
                        hp_colors = ['#00CC96', '#AB63FA', '#FFA15A', '#19D3F3', '#FF6692', '#B6E880', '#FF97FF', '#FECB52']
                        
                        # Show heat pump comparison table on left and chart on right
                        if all_hp_data:
                            hpi_col1, hpi_spacer, hpi_col2 = st.columns([0.35, 0.05, 0.60])
                            
                            with hpi_col1:
                                
                                # Create color mapping for heat pumps based on integration data order
                                hp_color_map = {}
                                for idx, hp_int in enumerate(all_hp_integration_data):
                                    hp_color_map[hp_int['name']] = hp_colors[idx % len(hp_colors)]
                                
                                # Only show available heat pumps, sorted by COP descending
                                available_data = []
                                
                                for hp in all_hp_data:
                                    if hp['available']:
                                        # Get color for this heat pump
                                        color = hp_color_map.get(hp['name'], '#999999')
                                        available_data.append({
                                            'Heat Pump': hp['name'],
                                            'COP': f"{hp['cop']:.2f}",
                                            'T_source (°C)': f"{hp['t_source']:.1f}",
                                            'T_sink (°C)': f"{hp['t_sink']:.1f}",
                                            'Q_source (kW)': f"{hp['q_source']:.1f}",
                                            'Q_sink (kW)': f"{hp['q_sink']:.1f}",
                                            'cop_value': hp['cop'],  # Keep for sorting
                                            'color': color
                                        })
                                
                                # Sort available heat pumps by COP (descending)
                                available_data.sort(key=lambda x: x['cop_value'], reverse=True)
                                
                                # Build table with dark-mode aware styling
                                table_html = (
                                    '<style>'
                                    '.hp-table{width:100%;border-collapse:collapse;font-size:12px;background-color:transparent;color:inherit;}'
                                    '.hp-table th{background-color:transparent;padding:8px;text-align:left;border:1px solid rgba(0,0,0,0.08);font-weight:bold;color:inherit;}'
                                    '.hp-table td{padding:8px;border:1px solid rgba(0,0,0,0.08);color:inherit;}'
                                    '.hp-table tr:nth-child(even){background-color:transparent;}'
                                    '@media (prefers-color-scheme: dark) {'
                                    '.hp-table th{border-color: rgba(255,255,255,0.06);}'
                                    '.hp-table td{border-color: rgba(255,255,255,0.06);}'
                                    '}'
                                    '</style>'
                                    '<table class="hp-table"><thead><tr><th>Heat Pump</th><th>COP</th><th>T_source (°C)</th><th>T_sink (°C)</th><th>Q_source (kW)</th><th>Q_sink (kW)</th></tr></thead><tbody>'
                                )
                                
                                for item in available_data:
                                    table_html += f'<tr><td>{item["Heat Pump"]}</td><td>{item["COP"]}</td><td>{item["T_source (°C)"]}</td><td>{item["T_sink (°C)"]}</td><td>{item["Q_source (kW)"]}</td><td>{item["Q_sink (kW)"]}</td></tr>'
                                
                                table_html += '</tbody></table>'
                                
                                st.markdown(table_html, unsafe_allow_html=True)
                                
                                # Show reasons for unavailable heat pumps
                                unavailable_reasons = [hp for hp in all_hp_data if not hp['available']]
                                if unavailable_reasons:
                                    with st.expander("ℹ️ Heat pumps that cannot be integrated"):
                                        for hp in unavailable_reasons:
                                            st.markdown(f"**{hp['name']}**: {hp['reason']}")
                            
                            with hpi_col2:
                                # Add multiselect for heat pump selection
                                selected_hps = st.multiselect(
                                    "Select Heat Pump Types to visualize:", 
                                    available_hp_names, 
                                    default=[available_hp_names[0]] if available_hp_names else [],
                                    key="hp_type_multiselect"
                                )
                                
                                # Filter integration data based on selection
                                hp_integration_data = [hp for hp in all_hp_integration_data if hp['name'] in selected_hps]
                                
                                # Create columns for legend and chart within hpi_col2
                                legend_col, chart_col = st.columns([0.15, 0.85])
                                
                                with legend_col:
                                    # Create custom legend with larger symbols
                                    st.markdown("**Legend**")
                                    legend_html = '<div style="margin-top: 10px; margin-left: 15px;">'
                                    for idx, hp_data in enumerate(hp_integration_data):
                                        color = hp_colors[idx % len(hp_colors)]
                                        hp_name = hp_data['name']
                                        legend_html += f'<div style="margin: 8px 0; display: flex; align-items: center; gap: 10px;"><span style="font-size: 36px; color: {color}; font-weight: bold; line-height: 1;">◆</span><span style="font-size: 15px;">{hp_name} - Source</span></div>'
                                        legend_html += f'<div style="margin: 8px 0; display: flex; align-items: center; gap: 10px;"><span style="font-size: 36px; color: {color}; font-weight: bold; line-height: 1;">◇</span><span style="font-size: 15px;">{hp_name} - Sink</span></div>'
                                    legend_html += '</div>'
                                    st.markdown(legend_html, unsafe_allow_html=True)
                                
                                with chart_col:
                                    # Create Plotly figure for HPI
                                    fig_hpi = go.Figure()
                                    
                                    # Plot GCC segments with colors (red for heating, blue for cooling)
                                    for i in range(len(gcc_H) - 1):
                                        # Determine color based on deltaH
                                        if i < len(hpi_analysis.pyPinch.heatCascade):
                                            dh = hpi_analysis.pyPinch.heatCascade[i]['deltaH']
                                            color = 'red' if dh > 0 else ('blue' if dh < 0 else 'gray')
                                        else:
                                            color = 'gray'
                                        
                                        fig_hpi.add_trace(go.Scatter(
                                            x=[gcc_H[i], gcc_H[i+1]],
                                            y=[gcc_T[i], gcc_T[i+1]],
                                            mode='lines+markers',
                                            line=dict(color=color, width=2),
                                            marker=dict(size=5),
                                            showlegend=False,
                                            hovertemplate='T: %{y:.1f}°C<br>H: %{x:.1f} kW<extra></extra>'
                                        ))
                                    
                                    # Add integration points for each selected heat pump
                                    for idx, hp_data in enumerate(hp_integration_data):
                                        color = hp_colors[idx % len(hp_colors)]
                                        hp_name = hp_data['name']
                                        int_temp = hp_data['int_temp']
                                        int_qsource = hp_data['int_qsource']
                                        int_qsink = hp_data['int_qsink']
                                        int_cop = hp_data['int_cop']
                                        t_sink = hp_data['t_sink']
                                        
                                        # Add diamond markers for source (evaporator)
                                        fig_hpi.add_trace(go.Scatter(
                                            x=[int_qsource],
                                            y=[int_temp],
                                            mode='markers',
                                            marker=dict(size=12, color=color, symbol='diamond', 
                                                      line=dict(width=2, color=color)),
                                            name=f'{hp_name} - Source',
                                            legendgroup=hp_name,
                                            hovertemplate=f'<b>{hp_name} - Source</b><br>T: {int_temp:.1f}°C<br>Q: {int_qsource:.1f} kW<br>COP: {int_cop:.2f}<extra></extra>'
                                        ))
                                        
                                        # Add diamond markers for sink (condenser)
                                        fig_hpi.add_trace(go.Scatter(
                                            x=[int_qsink],
                                            y=[t_sink],
                                            mode='markers',
                                            marker=dict(size=12, color=color, symbol='diamond-open',
                                                      line=dict(width=2, color=color)),
                                            name=f'{hp_name} - Sink',
                                            legendgroup=hp_name,
                                            hovertemplate=f'<b>{hp_name} - Sink</b><br>T: {t_sink:.1f}°C<br>Q: {int_qsink:.1f} kW<br>COP: {int_cop:.2f}<extra></extra>'
                                        ))
                                    
                                    # Add pinch line
                                    fig_hpi.add_hline(
                                        y=pinch.pinchTemperature,
                                        line_dash='dash',
                                        line_color='gray',
                                        annotation_text=f"Pinch: {pinch.pinchTemperature:.1f}°C",
                                        annotation_position='top right'
                                    )
                                    
                                    # Add zero enthalpy line
                                    fig_hpi.add_vline(x=0, line_color='black', line_width=1, opacity=0.3)
                                    
                                    fig_hpi.update_layout(
                                        title=dict(text='Heat Pump Integration - Grand Composite Curve', font=dict(size=14)),
                                        xaxis_title='Net ΔH (kW)',
                                        yaxis_title='Shifted Temperature (°C)',
                                        height=400,
                                        margin=dict(l=60, r=20, t=40, b=50),
                                        hovermode='closest',
                                        yaxis=dict(rangemode='tozero'),
                                        showlegend=False
                                    )
                                    
                                    st.plotly_chart(fig_hpi, key="hpi_chart")
                        else:
                            st.plotly_chart(fig_hpi, key="hpi_chart")
                            
                            # Show reasons for unavailable heat pumps
                            unavailable_reasons = [hp for hp in all_hp_data if not hp['available']]
                            if unavailable_reasons:
                                with st.expander("ℹ️ Heat pumps that cannot be integrated"):
                                    for hp in unavailable_reasons:
                                        st.markdown(f"**{hp['name']}**: {hp['reason']}")
                        
                    finally:
                        os.unlink(hpi_csv_path)
                        
                except ValueError as ve:
                    # This is our custom error for insufficient streams
                    pass  # Warning already shown above
                except IndexError:
                    st.warning("⚠️ Heat pump integration is not possible with the selected streams. Please select streams with both heating and cooling requirements that allow for heat pump integration.")
                except Exception as hpi_error:
                    st.warning(f"⚠️ Heat pump integration could not be performed with the selected streams: {str(hpi_error)}")
                
                # =====================================================
                # STATUS QUO VS PINCH ANALYSIS COMPARISON
                # =====================================================
                st.markdown("---")
                st.markdown("## Status Quo vs Pinch Analysis Comparison")
                
                # Check if we have current energy demands
                current_demands = st.session_state.get('current_energy_demands', [])
                has_demands = any(d['heat_demand'] > 0 or d['cooling_demand'] > 0 for d in current_demands)
                
                if not has_demands:
                    st.info("💡 Current energy demands are required to calculate the difference to status quo. Please configure energy demand sets in the 'Current energy demand set' section above.")
                else:
                    # Calculate total current demands
                    total_current_heat = sum(d['heat_demand'] for d in current_demands)
                    total_current_cooling = sum(d['cooling_demand'] for d in current_demands)
                    
                    # Check if pinch analysis was run (look for the results in the session or check if we have the data)
                    pinch_results_available = False
                    pinch_heat_min = 0
                    pinch_cooling_min = 0
                    
                    # Try to get pinch results from the analysis above
                    try:
                        if 'results' in locals() and results:
                            pinch_heat_min = results['hot_utility']
                            pinch_cooling_min = results['cold_utility']
                            pinch_results_available = True
                    except:
                        pinch_results_available = False
                    
                    if pinch_results_available:
                        # Calculate improvements (positive = savings, negative = additional need)
                        heat_improvement = total_current_heat - pinch_heat_min
                        cooling_improvement = total_current_cooling - pinch_cooling_min
                        
                        # Create comparison table
                        import pandas as pd
                        
                        comparison_data = {
                            'Energy Type': ['Heating Demand', 'Cooling Demand'],
                            'Current Status Quo (kW)': [f"{total_current_heat:.1f}", f"{total_current_cooling:.1f}"],
                            'Pinch Reduction (kW)': [f"-{pinch_heat_min:.1f}", f"-{pinch_cooling_min:.1f}"],
                            'Savings (kW)': [f"-{heat_improvement:.1f}", f"-{cooling_improvement:.1f}"],
                            'Percentage': [
                                f"-{abs(heat_improvement) / total_current_heat * 100:.1f}%" if total_current_heat > 0 else "N/A",
                                f"-{abs(cooling_improvement) / total_current_cooling * 100:.1f}%" if total_current_cooling > 0 else "N/A"
                            ]
                        }
                        
                        comparison_df = pd.DataFrame(comparison_data)
                        st.table(comparison_df)
                    else:
                        st.warning("⚠️ Pinch analysis results not available. Please ensure you have selected at least 2 streams with complete data to run the analysis.")
                
                # Notes section
                st.markdown("---")
                st.markdown("##### Notes")
                if 'pinch_notes' not in st.session_state:
                    st.session_state['pinch_notes'] = ""
                
                notes = st.text_area(
                    "Notes",
                    value=st.session_state['pinch_notes'],
                    height=100,
                    placeholder="Add notes about the pinch analysis here...",
                    key="pinch_notes_input",
                    label_visibility="collapsed"
                )
                st.session_state['pinch_notes'] = notes
                
                # More information expander
                with st.expander("More information"):
                    import pandas as pd
                    
                    
                    temps = results['temperatures']
                    pinch_streams = results['streams']
                    
                    if pinch_streams and temps:
                        fig_interval = go.Figure()
                        
                        num_streams = len(pinch_streams)
                        x_positions = [(i + 1) * 1.0 for i in range(num_streams)]
                        
                        # Draw horizontal temperature lines
                        for temperature in temps:
                            fig_interval.add_shape(
                                type="line",
                                x0=0, x1=num_streams + 1,
                                y0=temperature, y1=temperature,
                                line=dict(color="gray", width=1, dash="dot"),
                            )
                        
                        # Draw pinch temperature line
                        fig_interval.add_shape(
                            type="line",
                            x0=0, x1=num_streams + 1,
                            y0=results['pinch_temperature'], y1=results['pinch_temperature'],
                            line=dict(color="black", width=2, dash="dash"),
                        )
                        fig_interval.add_annotation(
                            x=num_streams + 0.5, y=results['pinch_temperature'],
                            text=f"Pinch: {results['pinch_temperature']:.1f}°C",
                            showarrow=False, font=dict(size=10),
                            xanchor='left'
                        )
                        
                        # Draw stream arrows
                        for i, stream in enumerate(pinch_streams):
                            ss = stream['ss']  # Shifted supply temp
                            st_temp = stream['st']  # Shifted target temp
                            stream_type = stream['type']
                            x_pos = x_positions[i]
                            
                            # Color based on stream type
                            color = 'red' if stream_type == 'HOT' else 'blue'
                            stream_name = streams_data[i]['name'] if i < len(streams_data) else f'Stream {i+1}'
                            
                            # Draw arrow as a line with annotation for arrowhead
                            fig_interval.add_trace(go.Scatter(
                                x=[x_pos, x_pos],
                                y=[ss, st_temp],
                                mode='lines',
                                line=dict(color=color, width=8),
                                hovertemplate=f"<b>{stream_name}</b><br>" +
                                             f"Type: {stream_type}<br>" +
                                             f"T_supply (shifted): {ss:.1f}°C<br>" +
                                             f"T_target (shifted): {st_temp:.1f}°C<br>" +
                                             f"CP: {stream['cp']:.2f} kW/K<extra></extra>",
                                showlegend=False
                            ))
                            
                            # Add arrowhead
                            fig_interval.add_annotation(
                                x=x_pos, y=st_temp,
                                ax=x_pos, ay=ss,
                                xref='x', yref='y',
                                axref='x', ayref='y',
                                showarrow=True,
                                arrowhead=2,
                                arrowsize=1.5,
                                arrowwidth=3,
                                arrowcolor=color
                            )
                            
                            # Stream label at top
                            label_y = max(ss, st_temp) + (max(temps) - min(temps)) * 0.03
                            fig_interval.add_annotation(
                                x=x_pos, y=label_y,
                                text=f"<b>S{i+1}</b>",
                                showarrow=False,
                                font=dict(size=11, color='white'),
                                bgcolor=color,
                                bordercolor='black',
                                borderwidth=1,
                                borderpad=3
                            )
                            
                            # CP value in middle
                            mid_y = (ss + st_temp) / 2
                            fig_interval.add_annotation(
                                x=x_pos, y=mid_y,
                                text=f"CP={stream['cp']:.1f}",
                                showarrow=False,
                                font=dict(size=9, color='white'),
                                textangle=-90
                            )
                        
                        fig_interval.update_layout(
                            title=dict(text='Shifted Temperature Interval Diagram', font=dict(size=14)),
                            xaxis=dict(
                                title='Streams',
                                showticklabels=False,
                                range=[0, num_streams + 1],
                                showgrid=False
                            ),
                            yaxis=dict(
                                title='Shifted Temperature S (°C)',
                                showgrid=True,
                                gridcolor='rgba(0,0,0,0.1)'
                            ),
                            height=400,
                            margin=dict(l=60, r=20, t=40, b=40),
                            hovermode='closest',
                            showlegend=False
                        )
                        
                        st.plotly_chart(fig_interval, width='stretch', key="interval_chart")
                
            except Exception as e:
                st.error(f"Error: {str(e)}")