Spaces:
Runtime error
Runtime error
Commit
·
3dc0572
1
Parent(s):
8411795
Update main.py
Browse files
main.py
CHANGED
@@ -1,440 +1,9 @@
|
|
1 |
-
import
|
2 |
import os
|
3 |
-
import streamlit as st
|
4 |
-
import numpy as np
|
5 |
-
import math
|
6 |
-
import fastf1
|
7 |
-
import pandas as pd
|
8 |
-
from fastapi import FastAPI
|
9 |
-
from fastapi.middleware.cors import CORSMiddleware
|
10 |
-
from fastapi.responses import FileResponse, HTMLResponse
|
11 |
|
|
|
12 |
|
13 |
-
|
|
|
14 |
|
15 |
-
import
|
16 |
-
import math
|
17 |
-
import numpy as np
|
18 |
-
import concurrent.futures
|
19 |
-
|
20 |
-
|
21 |
-
import available_data
|
22 |
-
|
23 |
-
app = FastAPI()
|
24 |
-
|
25 |
-
|
26 |
-
|
27 |
-
app.add_middleware(
|
28 |
-
CORSMiddleware,
|
29 |
-
allow_origins=["*"],
|
30 |
-
allow_credentials=True,
|
31 |
-
allow_methods=["*"],
|
32 |
-
allow_headers=["*"],
|
33 |
-
)
|
34 |
-
|
35 |
-
FASTF1_CACHE_DIR = os.environ['FASTF1_CACHE_DIR']
|
36 |
-
|
37 |
-
fastf1.Cache.enable_cache(FASTF1_CACHE_DIR)
|
38 |
-
|
39 |
-
|
40 |
-
def smooth_derivative(t_in, v_in):
|
41 |
-
|
42 |
-
#
|
43 |
-
# Function to compute a smooth estimation of a derivative.
|
44 |
-
# [REF: http://holoborodko.com/pavel/numerical-methods/numerical-derivative/smooth-low-noise-differentiators/]
|
45 |
-
#
|
46 |
-
|
47 |
-
# Configuration
|
48 |
-
#
|
49 |
-
# Derivative method: two options: 'smooth' or 'centered'. Smooth is more conservative
|
50 |
-
# but helps to supress the very noisy signals. 'centered' is more agressive but more noisy
|
51 |
-
method = "smooth"
|
52 |
-
|
53 |
-
t = t_in.copy()
|
54 |
-
v = v_in.copy()
|
55 |
-
|
56 |
-
# (0) Prepare inputs
|
57 |
-
|
58 |
-
# (0.1) Time needs to be transformed to seconds
|
59 |
-
try:
|
60 |
-
for i in range(0, t.size):
|
61 |
-
t.iloc[i] = t.iloc[i].total_seconds()
|
62 |
-
except:
|
63 |
-
pass
|
64 |
-
|
65 |
-
t = np.array(t)
|
66 |
-
v = np.array(v)
|
67 |
-
|
68 |
-
# (0.1) Assert they have the same size
|
69 |
-
assert t.size == v.size
|
70 |
-
|
71 |
-
# (0.2) Initialize output
|
72 |
-
dvdt = np.zeros(t.size)
|
73 |
-
|
74 |
-
# (1) Manually compute points out of the stencil
|
75 |
-
|
76 |
-
# (1.1) First point
|
77 |
-
dvdt[0] = (v[1] - v[0]) / (t[1] - t[0])
|
78 |
-
|
79 |
-
# (1.2) Second point
|
80 |
-
dvdt[1] = (v[2] - v[0]) / (t[2] - t[0])
|
81 |
-
|
82 |
-
# (1.3) Third point
|
83 |
-
dvdt[2] = (v[3] - v[1]) / (t[3] - t[1])
|
84 |
-
|
85 |
-
# (1.4) Last points
|
86 |
-
n = t.size
|
87 |
-
dvdt[n - 1] = (v[n - 1] - v[n - 2]) / (t[n - 1] - t[n - 2])
|
88 |
-
dvdt[n - 2] = (v[n - 1] - v[n - 3]) / (t[n - 1] - t[n - 3])
|
89 |
-
dvdt[n - 3] = (v[n - 2] - v[n - 4]) / (t[n - 2] - t[n - 4])
|
90 |
-
|
91 |
-
# (2) Compute the rest of the points
|
92 |
-
if method == "smooth":
|
93 |
-
c = [5.0 / 32.0, 4.0 / 32.0, 1.0 / 32.0]
|
94 |
-
for i in range(3, t.size - 3):
|
95 |
-
for j in range(1, 4):
|
96 |
-
dvdt[i] += (
|
97 |
-
2 * j * c[j - 1] * (v[i + j] - v[i - j]) /
|
98 |
-
(t[i + j] - t[i - j])
|
99 |
-
)
|
100 |
-
elif method == "centered":
|
101 |
-
for i in range(3, t.size - 2):
|
102 |
-
for j in range(1, 4):
|
103 |
-
dvdt[i] = (v[i + 1] - v[i - 1]) / (t[i + 1] - t[i - 1])
|
104 |
-
|
105 |
-
return dvdt
|
106 |
-
|
107 |
-
|
108 |
-
def truncated_remainder(dividend, divisor):
|
109 |
-
divided_number = dividend / divisor
|
110 |
-
divided_number = (
|
111 |
-
-int(-divided_number) if divided_number < 0 else int(divided_number)
|
112 |
-
)
|
113 |
-
|
114 |
-
remainder = dividend - divisor * divided_number
|
115 |
-
|
116 |
-
return remainder
|
117 |
-
|
118 |
-
|
119 |
-
def transform_to_pipi(input_angle):
|
120 |
-
pi = math.pi
|
121 |
-
revolutions = int((input_angle + np.sign(input_angle) * pi) / (2 * pi))
|
122 |
-
|
123 |
-
p1 = truncated_remainder(input_angle + np.sign(input_angle) * pi, 2 * pi)
|
124 |
-
p2 = (
|
125 |
-
np.sign(
|
126 |
-
np.sign(input_angle)
|
127 |
-
+ 2
|
128 |
-
* (
|
129 |
-
np.sign(
|
130 |
-
math.fabs(
|
131 |
-
(truncated_remainder(input_angle + pi, 2 * pi)) / (2 * pi)
|
132 |
-
)
|
133 |
-
)
|
134 |
-
- 1
|
135 |
-
)
|
136 |
-
)
|
137 |
-
) * pi
|
138 |
-
|
139 |
-
output_angle = p1 - p2
|
140 |
-
|
141 |
-
return output_angle, revolutions
|
142 |
-
|
143 |
-
|
144 |
-
def remove_acceleration_outliers(acc):
|
145 |
-
|
146 |
-
acc_threshold_g = 7.5
|
147 |
-
if math.fabs(acc[0]) > acc_threshold_g:
|
148 |
-
acc[0] = 0.0
|
149 |
-
|
150 |
-
for i in range(1, acc.size - 1):
|
151 |
-
if math.fabs(acc[i]) > acc_threshold_g:
|
152 |
-
acc[i] = acc[i - 1]
|
153 |
-
|
154 |
-
if math.fabs(acc[-1]) > acc_threshold_g:
|
155 |
-
acc[-1] = acc[-2]
|
156 |
-
|
157 |
-
return acc
|
158 |
-
|
159 |
-
|
160 |
-
def compute_accelerations(telemetry):
|
161 |
-
|
162 |
-
v = np.array(telemetry["Speed"]) / 3.6
|
163 |
-
lon_acc = smooth_derivative(telemetry["Time"], v) / 9.81
|
164 |
-
|
165 |
-
dx = smooth_derivative(telemetry["Distance"], telemetry["X"])
|
166 |
-
dy = smooth_derivative(telemetry["Distance"], telemetry["Y"])
|
167 |
-
|
168 |
-
theta = np.zeros(dx.size)
|
169 |
-
theta[0] = math.atan2(dy[0], dx[0])
|
170 |
-
for i in range(0, dx.size):
|
171 |
-
theta[i] = (
|
172 |
-
theta[i - 1] +
|
173 |
-
transform_to_pipi(math.atan2(dy[i], dx[i]) - theta[i - 1])[0]
|
174 |
-
)
|
175 |
-
|
176 |
-
kappa = smooth_derivative(telemetry["Distance"], theta)
|
177 |
-
lat_acc = v * v * kappa / 9.81
|
178 |
-
|
179 |
-
# Remove outliers
|
180 |
-
lon_acc = remove_acceleration_outliers(lon_acc)
|
181 |
-
lat_acc = remove_acceleration_outliers(lat_acc)
|
182 |
-
|
183 |
-
return np.round(lon_acc,2), np.round(lat_acc,2)
|
184 |
-
|
185 |
-
# @st.cache_data
|
186 |
-
@app.get("/wdc", response_model=None)
|
187 |
-
async def driver_standings() -> any:
|
188 |
-
YEAR = 2023 #datetime.datetime.now().year
|
189 |
-
df = pd.DataFrame(
|
190 |
-
pd.read_html(f"https://www.formula1.com/en/results.html/{YEAR}/drivers.html")[0]
|
191 |
-
)
|
192 |
-
df = df[["Driver", "PTS", "Car"]]
|
193 |
-
# reverse the order
|
194 |
-
df = df.sort_values(by="PTS", ascending=True)
|
195 |
-
|
196 |
-
# in Driver column only keep the last 3 characters
|
197 |
-
df["Driver"] = df["Driver"].str[:-5]
|
198 |
-
|
199 |
-
# add colors to the dataframe
|
200 |
-
car_colors = available_data.team_colors(YEAR)
|
201 |
-
df["fill"] = df["Car"].map(car_colors)
|
202 |
-
|
203 |
-
|
204 |
-
# remove rows where points is 0
|
205 |
-
df = df[df["PTS"] != 0]
|
206 |
-
df.reset_index(inplace=True, drop=True)
|
207 |
-
df.rename(columns={"PTS": "Points"}, inplace=True)
|
208 |
-
|
209 |
-
return {"WDC":df.to_dict("records")}
|
210 |
-
|
211 |
-
# @st.cache_data
|
212 |
-
@app.get("/", response_model=None)
|
213 |
-
async def root():
|
214 |
-
return HTMLResponse(
|
215 |
-
content="""<iframe src="https://tracinginsights-f1-analysis.hf.space" frameborder="0" style="width:100%; height:100%;" scrolling="yes" allowfullscreen:"yes"></iframe>""",
|
216 |
-
status_code=200)
|
217 |
-
|
218 |
-
# @st.cache_data
|
219 |
-
@app.get("/years", response_model=None)
|
220 |
-
async def years_available() -> any:
|
221 |
-
# make a list from 2018 to current year
|
222 |
-
current_year = datetime.datetime.now().year
|
223 |
-
years = list(range(2018, current_year+1))
|
224 |
-
# reverse the list to get the latest year first
|
225 |
-
years.reverse()
|
226 |
-
years = [{"label": str(year), "value": year} for year in years]
|
227 |
-
return {"years": years}
|
228 |
-
|
229 |
-
|
230 |
-
# format for events {"events":[{"label":"Saudi Arabian Grand Prix","value":2},{"label":"Bahrain Grand Prix","value":1},{"label":"Pre-Season Testing","value":"t1"}]}
|
231 |
-
|
232 |
-
# @st.cache_data
|
233 |
-
@app.get("/{year}", response_model=None)
|
234 |
-
async def events_available(year: int) -> any:
|
235 |
-
# get events available for a given year
|
236 |
-
data = available_data.LatestData(year)
|
237 |
-
events = data.get_events()
|
238 |
-
events = [{"label": event, "value": event} for i, event in enumerate(events)]
|
239 |
-
events.reverse()
|
240 |
-
|
241 |
-
return {"events": events}
|
242 |
-
|
243 |
-
# format for sessions {"sessions":[{"label":"FP1","value":"FP1"},{"label":"FP2","value":"FP2"},{"label":"FP3","value":"FP3"},{"label":"Qualifying","value":"Q"},{"label":"Race","value":"R"}]}
|
244 |
-
|
245 |
-
# @st.cache_data
|
246 |
-
@functools.cache
|
247 |
-
@app.get("/{year}/{event}", response_model=None)
|
248 |
-
async def sessions_available(year: int, event: str | int) -> any:
|
249 |
-
# get sessions available for a given year and event
|
250 |
-
data = available_data.LatestData(year)
|
251 |
-
sessions = data.get_sessions(event)
|
252 |
-
sessions = [{"label": session, "value": session} for session in sessions]
|
253 |
-
|
254 |
-
return {"sessions": sessions}
|
255 |
-
|
256 |
-
# format for drivers {"drivers":[{"color":"#fff500","label":"RIC","value":"RIC"},{"color":"#ff8700","label":"NOR","value":"NOR"},{"color":"#c00000","label":"VET","value":"VET"},{"color":"#0082fa","label":"LAT","value":"LAT"},{"color":"#787878","label":"GRO","value":"GRO"},{"color":"#ffffff","label":"GAS","value":"GAS"},{"color":"#f596c8","label":"STR","value":"STR"},{"color":"#787878","label":"MAG","value":"MAG"},{"color":"#0600ef","label":"ALB","value":"ALB"},{"color":"#ffffff","label":"KVY","value":"KVY"},{"color":"#fff500","label":"OCO","value":"OCO"},{"color":"#0600ef","label":"VER","value":"VER"},{"color":"#00d2be","label":"HAM","value":"HAM"},{"color":"#ff8700","label":"SAI","value":"SAI"},{"color":"#00d2be","label":"BOT","value":"BOT"},{"color":"#960000","label":"GIO","value":"GIO"}]}
|
257 |
-
|
258 |
-
# @st.cache_data
|
259 |
-
@functools.cache
|
260 |
-
@app.get("/{year}/{event}/{session}", response_model=None)
|
261 |
-
async def session_drivers(year: int, event: str | int, session: str) -> any:
|
262 |
-
|
263 |
-
# get drivers available for a given year, event and session
|
264 |
-
f1session = fastf1.get_session(year, event, session)
|
265 |
-
api_path = f1session.api_path
|
266 |
-
drivers_raw = fastf1.api.driver_info(api_path)
|
267 |
-
drivers = [{
|
268 |
-
"color": available_data.team_colors(year)[driver[1]['TeamName']],
|
269 |
-
"label": driver[1]['Tla'],
|
270 |
-
"value": driver[1]['Tla']
|
271 |
-
} for driver in drivers_raw.items()]
|
272 |
-
|
273 |
-
return {"drivers": drivers}
|
274 |
-
|
275 |
-
|
276 |
-
# format for chartData {"chartData":[{"lapnumber":1},{
|
277 |
-
# "VER":91.564,
|
278 |
-
# "VER_compound":"SOFT",
|
279 |
-
# "VER_compound_color":"#FF5733",
|
280 |
-
# "lapnumber":2
|
281 |
-
# },{"lapnumber":3},{"VER":90.494,"VER_compound":"SOFT","VER_compound_color":"#FF5733","lapnumber":4},{"lapnumber":5},{"VER":90.062,"VER_compound":"SOFT","VER_compound_color":"#FF5733","lapnumber":6},{"lapnumber":7},{"VER":89.815,"VER_compound":"SOFT","VER_compound_color":"#FF5733","lapnumber":8},{"VER":105.248,"VER_compound":"SOFT","VER_compound_color":"#FF5733","lapnumber":9},{"lapnumber":10},{"VER":89.79,"VER_compound":"SOFT","VER_compound_color":"#FF5733","lapnumber":11},{"VER":145.101,"VER_compound":"SOFT","VER_compound_color":"#FF5733","lapnumber":12},{"lapnumber":13},{"VER":89.662,"VER_compound":"SOFT","VER_compound_color":"#FF5733","lapnumber":14},{"lapnumber":15},{"VER":89.617,"VER_compound":"SOFT","VER_compound_color":"#FF5733","lapnumber":16},{"lapnumber":17},{"VER":140.717,"VER_compound":"SOFT","VER_compound_color":"#FF5733","lapnumber":18}]}
|
282 |
-
|
283 |
-
# @st.cache_data
|
284 |
-
@functools.cache
|
285 |
-
@app.get("/{year}/{event}/{session}/{driver}", response_model=None)
|
286 |
-
async def laps_data(year: int, event: str | int, session: str, driver: str) -> any:
|
287 |
-
|
288 |
-
|
289 |
-
# get drivers available for a given year, event and session
|
290 |
-
f1session = fastf1.get_session(year, event, session)
|
291 |
-
f1session.load(telemetry=False, weather=False, messages=False)
|
292 |
-
laps = f1session.laps
|
293 |
-
team_colors = available_data.team_colors(year)
|
294 |
-
# add team_colors dict to laps on Team column
|
295 |
-
|
296 |
-
drivers = laps.Driver.unique()
|
297 |
-
# for each driver in drivers, get the Team column from laps and get the color from team_colors dict
|
298 |
-
drivers = [{"color": team_colors[laps[laps.Driver ==
|
299 |
-
driver].Team.iloc[0]], "label": driver, "value": driver} for driver in drivers]
|
300 |
-
|
301 |
-
driver_laps = laps.pick_driver(driver)
|
302 |
-
driver_laps['LapTime'] = driver_laps['LapTime'].dt.total_seconds()
|
303 |
-
compound_colors = {
|
304 |
-
"SOFT": "#FF0000",
|
305 |
-
"MEDIUM": "#FFFF00",
|
306 |
-
"HARD": "#FFFFFF",
|
307 |
-
"INTERMEDIATE": "#00FF00",
|
308 |
-
"WET": "#088cd0",
|
309 |
-
|
310 |
-
}
|
311 |
-
|
312 |
-
driver_laps_data = []
|
313 |
-
|
314 |
-
for _, row in driver_laps.iterrows():
|
315 |
-
|
316 |
-
if row['LapTime'] > 0:
|
317 |
-
lap = {f"{driver}": row['LapTime'],
|
318 |
-
f"{driver}_compound": row['Compound'],
|
319 |
-
f"{driver}_compound_color": compound_colors[row['Compound']],
|
320 |
-
"lapnumber": row['LapNumber']}
|
321 |
-
else:
|
322 |
-
lap = {"lapnumber": row['LapNumber']}
|
323 |
-
|
324 |
-
driver_laps_data.append(lap)
|
325 |
-
|
326 |
-
return {"chartData": driver_laps_data}
|
327 |
-
|
328 |
-
|
329 |
-
# @st.cache_data
|
330 |
-
@functools.cache
|
331 |
-
@app.get("/{year}/{event}/{session}/{driver}/{lap_number}", response_model=None)
|
332 |
-
async def telemetry_data(year: int, event: str | int, session: str, driver: str, lap_number: int) -> any:
|
333 |
-
|
334 |
-
|
335 |
-
f1session = fastf1.get_session(year, event, session)
|
336 |
-
f1session.load(telemetry=True, weather=False, messages=False)
|
337 |
-
laps = f1session.laps
|
338 |
-
|
339 |
-
driver_laps = laps.pick_driver(driver)
|
340 |
-
driver_laps['LapTime'] = driver_laps['LapTime'].dt.total_seconds()
|
341 |
-
|
342 |
-
# get the telemetry for lap_number
|
343 |
-
selected_lap = driver_laps[driver_laps.LapNumber == lap_number]
|
344 |
-
|
345 |
-
telemetry = selected_lap.get_telemetry()
|
346 |
-
|
347 |
-
lon_acc, lat_acc = compute_accelerations(telemetry)
|
348 |
-
telemetry["lon_acc"] = lon_acc
|
349 |
-
telemetry["lat_acc"] = lat_acc
|
350 |
-
|
351 |
-
telemetry['Time'] = telemetry['Time'].dt.total_seconds()
|
352 |
-
|
353 |
-
laptime = selected_lap.LapTime.values[0]
|
354 |
-
data_key = f"{driver} - Lap {int(lap_number)} - {year} {session} [{int(laptime//60)}:{laptime%60}]"
|
355 |
-
|
356 |
-
telemetry['DRS'] = telemetry['DRS'].apply(lambda x: 1 if x in [10,12,14] else 0)
|
357 |
-
|
358 |
-
brake_tel = []
|
359 |
-
drs_tel = []
|
360 |
-
gear_tel = []
|
361 |
-
rpm_tel = []
|
362 |
-
speed_tel = []
|
363 |
-
throttle_tel = []
|
364 |
-
time_tel = []
|
365 |
-
track_map = []
|
366 |
-
lon_acc_tel = []
|
367 |
-
lat_acc_tel = []
|
368 |
-
|
369 |
-
for _, row in telemetry.iterrows():
|
370 |
-
|
371 |
-
brake = {"x": row['Distance'],
|
372 |
-
"y": row['Brake'],
|
373 |
-
}
|
374 |
-
brake_tel.append(brake)
|
375 |
-
|
376 |
-
drs = {"x": row['Distance'],
|
377 |
-
"y": row['DRS'],
|
378 |
-
}
|
379 |
-
drs_tel.append(drs)
|
380 |
-
|
381 |
-
gear = {"x": row['Distance'],
|
382 |
-
"y": row['nGear'],
|
383 |
-
}
|
384 |
-
gear_tel.append(gear)
|
385 |
-
|
386 |
-
rpm = {"x": row['Distance'],
|
387 |
-
"y": row['RPM'],
|
388 |
-
}
|
389 |
-
rpm_tel.append(rpm)
|
390 |
-
|
391 |
-
speed = {"x": row['Distance'],
|
392 |
-
"y": row['Speed'],
|
393 |
-
}
|
394 |
-
speed_tel.append(speed)
|
395 |
-
|
396 |
-
throttle = {"x": row['Distance'],
|
397 |
-
"y": row['Throttle'],
|
398 |
-
}
|
399 |
-
throttle_tel.append(throttle)
|
400 |
-
|
401 |
-
time = {"x": row['Distance'],
|
402 |
-
"y": row['Time'],
|
403 |
-
}
|
404 |
-
time_tel.append(time)
|
405 |
-
|
406 |
-
lon_acc = {"x": row['Distance'],
|
407 |
-
"y": row['lon_acc'],
|
408 |
-
}
|
409 |
-
lon_acc_tel.append(lon_acc)
|
410 |
-
|
411 |
-
lat_acc = {"x": row['Distance'],
|
412 |
-
"y": row['lat_acc'],
|
413 |
-
}
|
414 |
-
lat_acc_tel.append(lat_acc)
|
415 |
-
|
416 |
-
track = {"x": row['X'],
|
417 |
-
"y": row['Y'],
|
418 |
-
}
|
419 |
-
track_map.append(track)
|
420 |
-
|
421 |
-
|
422 |
-
|
423 |
-
telemetry_data = {
|
424 |
-
"telemetryData":{
|
425 |
-
"brake": brake_tel,
|
426 |
-
"dataKey": data_key,
|
427 |
-
"drs": drs_tel,
|
428 |
-
"gear": gear_tel,
|
429 |
-
"rpm": rpm_tel,
|
430 |
-
"speed": speed_tel,
|
431 |
-
"throttle": throttle_tel,
|
432 |
-
"time": time_tel,
|
433 |
-
"lon_acc": lon_acc_tel,
|
434 |
-
"lat_acc": lat_acc_tel,
|
435 |
-
"trackMap": track_map,
|
436 |
-
}
|
437 |
-
}
|
438 |
-
|
439 |
-
return telemetry_data
|
440 |
-
|
|
|
1 |
+
from git import Repo
|
2 |
import os
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
3 |
|
4 |
+
GITHUB_PAT = os.environ['GITHUB']
|
5 |
|
6 |
+
if not os.path.exists('repo_directory'):
|
7 |
+
Repo.clone_from(f'https://tracinginsights:{GITHUB_PAT}@github.com/TracingInsights/fastf1api.git', 'repo_directory' )
|
8 |
|
9 |
+
from repo_directory import *
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|