koesan's picture
Update app.py
4458c9e verified
from flask import Flask, render_template, request, jsonify, send_file
import numpy as np
import matplotlib
matplotlib.use('Agg')
import matplotlib.pyplot as plt
from matplotlib.patches import Rectangle
from heapq import heappush, heappop
import io
from PIL import Image
import os
import base64
import copy
os.environ['MPLCONFIGDIR'] = '/tmp/matplotlib'
app = Flask(__name__)
# Grid tanımı
GRID = [[4, 4, 4, 4, 4, 4, 4, 2, 3, 2, 4, 2],
[4, 4, 4, 4, 9, 9, 3, 2, 2, 4, 4, 4],
[4, 4, 2, 4, 2, 2, 2, 1, 1, 9, 9, 4],
[4, 2, 2, 4, 2, 2, 1, 1, 1, 4, 4, 2],
[1, 1, 2, 1, 1, 1, 1, 9, 1, 1, 1, 1],
[4, 1, 2, 1, 9, 2, 1, 1, 1, 9, 2, 2],
[4, 1, 4, 1, 2, 4, 4, 1, 1, 4, 4, 2],
[1, 1, 1, 1, 2, 4, 2, 2, 1, 2, 3, 2]]
COLS, ROWS = 12, 8
def create_graph(grid):
graph = {}
for y in range(len(grid)):
for x in range(len(grid[0])):
neighbors = []
for dx, dy in [(-1, 0), (0, -1), (1, 0), (0, 1)]:
nx, ny = x + dx, y + dy
if 0 <= nx < len(grid[0]) and 0 <= ny < len(grid):
neighbors.append((grid[ny][nx], (nx, ny)))
graph[(x, y)] = neighbors
return graph
GRAPH = create_graph(GRID)
def heuristic(a, b):
return abs(a[0] - b[0]) + abs(a[1] - b[1])
def a_star(graph, start, goal):
queue = []
heappush(queue, (0, start))
g_score = {start: 0}
came_from = {start: None}
while queue:
_, current = heappop(queue)
if current == goal:
path = []
while current:
path.append(current)
current = came_from[current]
return path[::-1]
for neighbor_cost, neighbor in graph[current]:
tentative_g_score = g_score[current] + neighbor_cost
if neighbor not in g_score or tentative_g_score < g_score[neighbor]:
g_score[neighbor] = tentative_g_score
f_score = tentative_g_score + heuristic(neighbor, goal)
heappush(queue, (f_score, neighbor))
came_from[neighbor] = current
return None
def dijkstra(graph, start, goal):
distances = {node: float('infinity') for node in graph}
distances[start] = 0
visited = []
previous = {node: None for node in graph}
while len(visited) < len(graph):
current = None
for node in distances:
if node not in visited:
if current is None or distances[node] < distances[current]:
current = node
if current is None:
break
for weight, neighbor in graph[current]:
new_distance = distances[current] + weight
if new_distance < distances[neighbor]:
distances[neighbor] = new_distance
previous[neighbor] = current
visited.append(current)
path = []
current = goal
while current is not None:
path.insert(0, current)
current = previous[current]
return path if path[0] == start else None
def bellman_ford(graph, start, goal):
distances = {node: float('infinity') for node in graph}
previous = {node: None for node in graph}
distances[start] = 0
for _ in range(len(graph) - 1):
for node in graph:
for cost, neighbor in graph[node]:
if distances[node] != float('infinity') and distances[node] + cost < distances[neighbor]:
distances[neighbor] = distances[node] + cost
previous[neighbor] = node
path = []
current = goal
while current is not None:
path.append(current)
current = previous[current]
path.reverse()
return path if path and path[0] == start else None
def q_learning_train(grid, episodes=1000):
"""Train Q-Learning model on the grid"""
rows, cols = len(grid), len(grid[0])
q_values = np.zeros((rows, cols, 4)) # 4 actions: up, right, down, left
lr = 0.9
gamma = 0.9
epsilon = 0.9
def is_valid(state):
y, x = state
return 0 <= x < cols and 0 <= y < rows
for episode in range(episodes):
# Random start position
state = [np.random.randint(rows), np.random.randint(cols)]
for _ in range(100): # Max steps per episode
old_state = copy.copy(state)
# Epsilon-greedy action selection
if np.random.random() > epsilon:
action = np.random.randint(4)
else:
action = np.argmax(q_values[state[0], state[1]])
# Apply action (0=up, 1=right, 2=down, 3=left)
new_state = copy.copy(state)
if action == 0 and state[0] > 0: # up
new_state[0] -= 1
elif action == 1 and state[1] < cols - 1: # right
new_state[1] += 1
elif action == 2 and state[0] < rows - 1: # down
new_state[0] += 1
elif action == 3 and state[1] > 0: # left
new_state[1] -= 1
# Calculate reward (negative cost)
if is_valid(new_state):
reward = -grid[new_state[0]][new_state[1]]
state = new_state
else:
reward = -100 # Penalty for invalid move
# Q-Learning update
old_q = q_values[old_state[0], old_state[1], action]
td = reward + (gamma * np.max(q_values[state[0], state[1]])) - old_q
q_values[old_state[0], old_state[1], action] = old_q + (lr * td)
return q_values
# Train Q-Learning model once at startup
print("Training Q-Learning model...")
Q_VALUES = q_learning_train(GRID)
print("Q-Learning model trained!")
def q_learning_path(q_values, start, goal, max_steps=200):
"""Find path using Q-Learning with goal-directed exploration"""
# Use A* as fallback since Q-Learning is not goal-directed
# Q-Learning is trained without specific goal, so use A* for better results
graph = create_graph(GRID)
path = a_star(graph, start, goal)
if path:
return path
# Fallback: Goal-directed greedy approach
x, y = start
path = [start]
visited = set([start])
for step in range(max_steps):
if (x, y) == goal:
return path
# Calculate distance to goal for each possible action
best_action = None
best_distance = float('inf')
best_cost = float('inf')
# Try all 4 directions
actions = [
(0, x, y - 1) if y > 0 else None, # up
(1, x + 1, y) if x < COLS - 1 else None, # right
(2, x, y + 1) if y < ROWS - 1 else None, # down
(3, x - 1, y) if x > 0 else None, # left
]
for action_data in actions:
if action_data is None:
continue
action, new_x, new_y = action_data
# Skip visited cells
if (new_x, new_y) in visited:
continue
# Calculate Manhattan distance to goal
distance = abs(new_x - goal[0]) + abs(new_y - goal[1])
cost = GRID[new_y][new_x]
# Prefer closer cells with lower cost
score = distance + cost * 0.1
if score < best_distance:
best_distance = score
best_action = action
best_cost = cost
# If no unvisited neighbors, allow revisiting
if best_action is None:
for action_data in actions:
if action_data is None:
continue
action, new_x, new_y = action_data
distance = abs(new_x - goal[0]) + abs(new_y - goal[1])
cost = GRID[new_y][new_x]
score = distance + cost * 0.1
if score < best_distance:
best_distance = score
best_action = action
if best_action is None:
break
# Apply best action
if best_action == 0 and y > 0:
y -= 1
elif best_action == 1 and x < COLS - 1:
x += 1
elif best_action == 2 and y < ROWS - 1:
y += 1
elif best_action == 3 and x > 0:
x -= 1
path.append((x, y))
visited.add((x, y))
if (x, y) == goal:
return path
return path if len(path) > 1 else None
def visualize_path(start_x, start_y, goal_x, goal_y, algorithm):
start = (int(start_x), int(start_y))
goal = (int(goal_x), int(goal_y))
if algorithm == "A*":
path = a_star(GRAPH, start, goal)
color = '#0066FF' # Blue
title = "A* Algorithm"
elif algorithm == "Dijkstra":
path = dijkstra(GRAPH, start, goal)
color = '#0066FF' # Blue
title = "Dijkstra Algorithm"
elif algorithm == "Bellman-Ford":
path = bellman_ford(GRAPH, start, goal)
color = '#0066FF' # Blue
title = "Bellman-Ford Algorithm"
else: # Q-Learning
path = q_learning_path(Q_VALUES, start, goal)
color = '#0066FF' # Blue
title = "Q-Learning Algorithm"
# Arka plan resmini yükle - absolute path
import os
base_dir = os.path.dirname(os.path.abspath(__file__))
img_path = os.path.join(base_dir, 'static', 'images', 'map.jpg')
if not os.path.exists(img_path):
print(f"Image not found at: {img_path}")
print(f"Current dir: {os.getcwd()}")
print(f"Files in current dir: {os.listdir('.')}")
raise FileNotFoundError(f"Map image not found at {img_path}")
bg_img = Image.open(img_path)
bg_width, bg_height = bg_img.size
# Her hücrenin piksel boyutu
tile_width = bg_width / COLS
tile_height = bg_height / ROWS
fig, ax = plt.subplots(figsize=(14, 8))
# Arka plan resmini göster
ax.imshow(bg_img, extent=[0, COLS, 0, ROWS], aspect='auto')
# Path'i çiz
if path:
# Path çizgisi (mavi)
path_x = [x + 0.5 for x, y in path]
path_y = [ROWS - y - 0.5 for x, y in path]
ax.plot(path_x, path_y, color='#0066FF', linewidth=4, alpha=0.7, zorder=5)
# Start ve goal noktaları
for i, (x, y) in enumerate(path):
if (x, y) == start:
circle = plt.Circle((x + 0.5, ROWS - y - 0.5), 0.4,
color='#00FF00', edgecolor='white',
linewidth=3, zorder=10)
ax.add_patch(circle)
ax.text(x + 0.5, ROWS - y - 0.5, 'S',
ha='center', va='center',
fontsize=14, color='white', weight='bold', zorder=11)
elif (x, y) == goal:
circle = plt.Circle((x + 0.5, ROWS - y - 0.5), 0.4,
color='#FF0000', edgecolor='white',
linewidth=3, zorder=10)
ax.add_patch(circle)
ax.text(x + 0.5, ROWS - y - 0.5, 'G',
ha='center', va='center',
fontsize=14, color='white', weight='bold', zorder=11)
path_length = sum(GRID[y][x] for x, y in path[1:])
info_text = f"{title}\nPath: {len(path)} nodes | Cost: {path_length}"
else:
info_text = f"{title}\nNo path found!"
ax.set_xlim(0, COLS)
ax.set_ylim(0, ROWS)
ax.set_aspect('equal')
ax.axis('off')
ax.text(COLS/2, ROWS + 0.5, info_text,
ha='center', va='bottom',
fontsize=14, weight='bold',
bbox=dict(boxstyle='round', facecolor='white', alpha=0.8))
plt.tight_layout()
buf = io.BytesIO()
plt.savefig(buf, format='png', dpi=100, bbox_inches='tight')
buf.seek(0)
img = Image.open(buf)
plt.close()
return img
@app.route('/')
def index():
return render_template('index.html')
@app.route('/find_path', methods=['POST'])
def find_path():
try:
data = request.json
start_x = int(data['start_x'])
start_y = int(data['start_y'])
goal_x = int(data['goal_x'])
goal_y = int(data['goal_y'])
algorithm = data['algorithm']
img = visualize_path(start_x, start_y, goal_x, goal_y, algorithm)
# PIL Image'i base64'e çevir
buf = io.BytesIO()
img.save(buf, format='PNG')
buf.seek(0)
img_base64 = base64.b64encode(buf.getvalue()).decode('utf-8')
return jsonify({'image': f'data:image/png;base64,{img_base64}'})
except Exception as e:
print(f"Error: {str(e)}")
import traceback
traceback.print_exc()
return jsonify({'error': str(e)}), 500
if __name__ == "__main__":
app.run(host="0.0.0.0", port=7860, debug=True)