#!/usr/bin/env python
# coding: utf-8

# In[1]:


get_ipython().run_cell_magic('time', '', '%matplotlib inline\nfrom new_import import *\n')


# In[2]:


get_ipython().run_cell_magic('time', '', '# Dask gateway\ncluster, client = notebook_utils.initialize_dask(use_gateway=True, workers=(1,4))\ndc = datacube.Datacube()\n\n# Configure s3 access\nconfigure_s3_access(aws_unsigned=False, requester_pays=True, client=client)\n\nclient\n')


# In[3]:


## cấu hình thời gian lấy ảnh và tọa độ
# date_range = ('2022-09-01', '2023-10-01')
# longtitude_range = (105.86575, 105.94120)
# latitude_range = (9.65070, 9.69850)
date_range = ('2022-09-01', '2023-10-01')
longtitude_range = (105.5, 106.4)
latitude_range = (9.2, 10.0) 


# In[4]:


## truy vấn ảnh vệ tinh sen2
data = load_data(dc, date_range, longtitude_range, latitude_range)
notebook_utils.heading(notebook_utils.xarray_object_size(data))
display(data)


# In[5]:


# Specify the start and end times 
min_date = '2022-09-01' # Thời gian bắt đầu lấy data cho quá trình train
max_date = '2023-10-01' # Thời gian kết thúc lấy data cho quá trình train
# Just do 1 month for testing
# max_date = '2022-10-01' # Thời gian kết thúc lấy data cho quá trình train

# Specify a spatail region to search using latitude/longitude cooridinates
min_longitude, max_longitude = (105.5, 106.4)
min_latitude, max_latitude = (9.2, 10.0)

# Specify the product. In this case we want to use Sentinel-2 Level-2A data
product = 's2_l2a'

# Construct the search query dictionary
query = {
    'product': product,                     # Product name
    'x': (min_longitude, max_longitude),    # "x" axis bounds
    'y': (min_latitude, max_latitude),      # "y" axis bounds
    'time': (min_date, max_date),           # Any parsable date strings
}


# In[6]:


# Most common CRS
native_crs = notebook_utils.mostcommon_crs(dc, query)
print(f'Most common native CRS: {native_crs}')


# In[7]:


# Specify the spectral band measurements we want to use for a classification algorithm
measurements = ['red', 'nir', 'scl']

load_params = {
    'measurements': measurements,                   # Selected measurement or alias names
    'output_crs': native_crs,                       # Target EPSG code
    'resolution': (-10, 10),                        # Target resolution
    'group_by': 'solar_day',                        # Scene grouping
    'dask_chunks': {'x': 2048, 'y': 2048},          # Dask chunks
}


# In[8]:


get_ipython().run_cell_magic('time', '', '# The replacement "dc.load()" function for this product\ndata = load_s2l2a_with_offset(\n    dc,\n    query | load_params   # Combine the two dicts that contain our search and load parameters\n)\n\n# This line prints the total size of the dataset hat was loaded\nnotebook_utils.heading(notebook_utils.xarray_object_size(data))\n\ndisplay(data)\n')


# In[9]:


# %%time
# # Tiến hành loại bỏ các vị trí bị mây ảnh hưởng
# result = mask_clean(data)
# progress(result)


# In[10]:


# Tiến hành tính toán NDVI
ds1 = calculate_indices(result, index='NDVI', satellite_mission='s2')
ndvi = ds1["NDVI"]
display(ndvi)


# In[11]:


get_ipython().run_cell_magic('time', '', "## tính ndvi theo tháng\naverage_ndvi = ndvi.resample(time='1M').mean().persist()\nprogress(average_ndvi)\n")


# In[12]:


# compute average_ndvi
average_ndvi = average_ndvi.compute()


# In[13]:


# cấu hình vh vv file
# name_vh = "ThuanHoa/ThuanHoa_VH.tif"
# name_vv = "ThuanHoa/ThuanHoa_VV.tif"

# load dữ liệu sen1
bbox = [105.5, 9.2, 106.4, 10.0]
time_range = '2022-09-01/2023-10-01'
# dsvh, dsvv = load_sen1(bbox, time_range)

name_vh = "vh-0922_0923-full_ST.tif"
name_vv = "vv-0922_0923-full_ST.tif"

if not os.path.exists(name_vh):
    get_ipython().system('aws s3 cp s3://easi-asia-dc-data/staging/ctu/sentinel-1/vh-0922_0923-full_ST.tif vh-0922_0923-full_ST.tif')
if not os.path.exists(name_vv):
    get_ipython().system('aws s3 cp s3://easi-asia-dc-data/staging/ctu/sentinel-1/vv-0922_0923-full_ST.tif vv-0922_0923-full_ST.tif')
    
bbox = [105.5, 9.2, 106.4, 10.0]
time_range = '2022-09-01/2023-10-01'
dsvh, dsvv = load_sen1(bbox, time_range)


# In[27]:


from sklearn.preprocessing import PolynomialFeatures
from sklearn.linear_model import LinearRegression
from sklearn.ensemble import RandomForestRegressor


# In[28]:


average_ndvi = average_ndvi[:, :7680, :8687]
mask = ~np.isnan(average_ndvi)
print(average_ndvi.shape)
print(dsvh.shape)
print(dsvv.shape)
print(mask.shape)
X_train = np.stack([dsvh.values[mask], dsvv.values[mask]], axis=1)
y_train = average_ndvi.values[mask]


# In[29]:


model = LinearRegression()
model.fit(X_train, y_train)


# In[30]:


X_pred = np.stack([dsvh.values[~mask], dsvv.values[~mask]], axis=1)
average_ndvi.values[~mask] = model.predict(X_pred)


# In[31]:


average_ndvi_filled = xr.DataArray(average_ndvi, dims=average_ndvi.dims)


# In[32]:


plt.imshow(average_ndvi_filled.isel(time=6))


# In[65]:


plt.imshow(average_ndvi.isel(time=6))


# In[33]:


train_path = "train/ST_training data_updated_1130points.shp"


# In[34]:


train = load_train_data(train_path)


# In[37]:


datasets = get_data_sen1_and_sen2(train, average_ndvi_filled, dsvh, dsvv)


# In[39]:


# cấu hình nhãn dữ liệu
label_mapping = {
    "Lua tom": "0",
    "Lua": "1",
    "CHN": "2",
    "CLN": "3",
    "TS": "4",
    "Song": "5",
    "Dat xay dung": "6",
    "Rung": "7"
}

# chia tập dữ liệu train, val, test
X_train, X_val, X_test, y_train, y_val, y_test = split_train_data(train, label_mapping, datasets)


# In[40]:


# Huấn luyện mô hình
grid_search = train_with_rf(X_train, X_val, y_train, y_val)


# In[41]:


# kiểm tra độ chính xác với tập test
y_pred_test = grid_search.predict(X_test)
test_accuracy = accuracy_score(y_test, y_pred_test)
print(f"Accuracy for test data {round(test_accuracy, 2)*100} %")


# In[42]:


# Lưu mô hình huấn luyện
save_model("model_new.joblib", grid_search)


# In[43]:


# đóng client, cluster
client.close()
cluster.close()


# In[ ]: