Compare commits

...

10 Commits

Author SHA1 Message Date
ae859e0db4 Put content in a container
Signed-off-by: Emmy D'Anello <ynerant@emy.lu>
2023-04-25 15:14:57 +02:00
4332862419 Optimize computation
Signed-off-by: Emmy D'Anello <ynerant@emy.lu>
2023-04-25 15:05:23 +02:00
2004155e51 Multiple changes
Signed-off-by: Emmy D'Anello <ynerant@emy.lu>
2023-04-25 11:54:11 +02:00
089af1db75 Remove progress bar while searching routes 2023-02-19 18:01:16 +01:00
0f61b3bece Replace loading text by a spinner 2023-02-19 17:43:36 +01:00
06f836e13f Use local bootstrap5 files 2023-02-13 13:19:14 +01:00
9fbf4fb172 Add option to update dataset silently 2023-02-13 13:12:39 +01:00
4f326626bf Better display with accordions 2023-02-13 13:02:34 +01:00
c2b7664375 Add features to see all reachable cities 2023-02-13 12:39:42 +01:00
ba3bef3d27 Queue routes when necessary 2023-02-13 12:08:24 +01:00
3 changed files with 268 additions and 141 deletions

305
app.py
View File

@ -1,11 +1,13 @@
#!/usr/bin/env python3
from contextlib import nullcontext
import csv
from datetime import date, datetime, time, timedelta
import os
import json
from pytz import timezone
import requests
from time import sleep
import click
from flask import Flask, render_template
@ -18,13 +20,11 @@ from tqdm import tqdm
import config
app = Flask(__name__)
cli = AppGroup('tgvmax', help="Manage the TGVMax dataset.")
app.cli.add_command(cli)
app.config |= config.FLASK_CONFIG
db = SQLAlchemy(app)
@ -61,53 +61,58 @@ class RouteQueue(db.Model):
@cli.command("update-dataset")
def update_dataset():
@click.option('--verbose', '-v', is_flag=True, help="Display errors.")
def update_dataset(verbose: bool = False):
"""
Query the latest version of the SNCF OpenData dataset, as a CSV file.
"""
try:
resp = requests.get('https://ressources.data.sncf.com/explore/dataset/tgvmax/information/')
content = resp.content.decode().split('<script type="application/ld+json">')[1].split('</script>')[0].strip()
content = content.replace('\r', '')
content = content.replace('" \n', '" \\n')
content = content.replace('.\n', '.\\n')
content = content.replace('\n\n \nLa', '\\n\\n \\nLa')
content = content.replace('\n"', '\\n"')
resp = requests.get('https://ressources.data.sncf.com/explore/dataset/tgvmax/information/')
content = resp.content.decode().split('<script type="application/ld+json">')[1].split('</script>')[0].strip()
content = content.replace('\r', '')
content = content.replace('" \n', '" \\n')
content = content.replace('.\n', '.\\n')
content = content.replace('\n\n \nLa', '\\n\\n \\nLa')
content = content.replace('\n"', '\\n"')
info = json.loads(content)
modified_date = datetime.fromisoformat(info['dateModified'])
info = json.loads(content)
modified_date = datetime.fromisoformat(info['dateModified'])
utc = timezone('UTC')
last_modified = datetime.utcfromtimestamp(os.path.getmtime('tgvmax.csv')).replace(tzinfo=utc) if os.path.isfile('tgvmax.csv') else datetime(1, 1, 1, tzinfo=utc)
utc = timezone('UTC')
last_modified = datetime.utcfromtimestamp(os.path.getmtime('tgvmax.csv')).replace(tzinfo=utc) if os.path.isfile(
'tgvmax.csv') else datetime(1, 1, 1, tzinfo=utc)
if last_modified < modified_date:
if last_modified < modified_date:
if verbose:
print("Updating tgvmax.csv…")
with requests.get(info['distribution'][0]['contentUrl'], stream=True) as resp:
resp.raise_for_status()
with open('tgvmax.csv', 'wb') as f:
with tqdm(unit='io', unit_scale=True) as t:
for chunk in resp.iter_content(chunk_size=512 * 1024):
if chunk:
f.write(chunk)
with requests.get(info['distribution'][0]['contentUrl'], stream=True) as resp:
resp.raise_for_status()
with open('tgvmax.csv', 'wb') as f:
with tqdm(unit='io', unit_scale=True) if verbose else nullcontext() as t:
for chunk in resp.iter_content(chunk_size=512 * 1024):
if chunk:
f.write(chunk)
if verbose:
t.update(len(chunk))
os.utime('tgvmax.csv', (modified_date.timestamp(), modified_date.timestamp()))
os.utime('tgvmax.csv', (modified_date.timestamp(), modified_date.timestamp()))
if verbose:
print("Done")
print("Last modification:", modified_date)
except Exception as e:
print("An error occured while updating tgvmax.csv")
print(e)
else:
if verbose:
print("Last modification:", modified_date)
exit(2)
@cli.command("parse-csv")
@click.option('-F', '--flush', type=bool, is_flag=True, help="Flush the database before filling it.")
def parse_trains(flush: bool = False):
@click.option('--verbose', '-v', is_flag=True, help="Display errors.")
def parse_trains(flush: bool = False, verbose: bool = False):
"""
Parse the CSV file and store it to the database.
"""
if flush:
print("Flush database…")
if verbose:
print("Flush database…")
db.session.query(Train).delete()
last_modification = datetime.utcfromtimestamp(os.path.getmtime('tgvmax.csv')).replace(tzinfo=timezone('UTC'))
@ -115,8 +120,11 @@ def parse_trains(flush: bool = False):
with open('tgvmax.csv') as f:
first_line = True
already_seen = set()
for line in tqdm(csv.reader(f, delimiter=';')):
already_updated = set(x[0] for x in db.session.query(Train).filter(Train.last_modification > last_modification)\
.values(Train.id))
for line in (tqdm if verbose else lambda x: x)(csv.reader(f, delimiter=';')):
if first_line:
# Skip first line
first_line = False
continue
@ -129,6 +137,10 @@ def parse_trains(flush: bool = False):
print("Duplicate:", train_id)
continue
if train_id in already_updated:
# Already updated by the simulator
continue
train = Train(
id=train_id,
day=date.fromisoformat(line[0]),
@ -156,59 +168,69 @@ def parse_trains(flush: bool = False):
db.session.commit()
def find_routes(day, origin, destination):
trains = db.session.query(Train).filter_by(day=day, tgvmax=True).all()
def find_routes(day: date | datetime, origin: str, destination: str | None,
verbose: bool = False, min_dep: time = time(0, 0),
explored: dict | None = None):
if isinstance(day, datetime):
day = day.date()
if explored is None:
explored = {}
if origin not in explored:
explored[origin] = (min_dep, None)
valid_routes = []
max_dep = time(23, 59)
else:
max_dep, valid_routes = explored[origin]
if max_dep < min_dep:
# Already parsed these trains
return {destination: valid_routes}
explored[origin] = min_dep, None
trains = db.session.query(Train).filter_by(day=day, tgvmax=True, orig=origin)\
.filter(Train.dep >= min_dep, Train.dep < max_dep).all()
if not trains:
# No train in the requested interval
explored[origin] = (min_dep, valid_routes)
return {destination: valid_routes}
trains.sort(key=lambda train: train.dep)
db.session.commit()
explore = []
per_arr_explore = {}
valid_routes = []
for train in tqdm(trains):
if train.orig == origin:
it = [train]
if train.dest == destination:
# We hope that we have a direct train
valid_routes.append(it)
else:
explore.append(it)
per_arr_explore.setdefault(train.dest, [])
per_arr_explore[train.dest].append(it)
continue
for it in list(per_arr_explore.get(train.orig, [])):
if any(train.dest == tr.dest or train.dest == origin for tr in it):
# Avoid loops
for train in (t := tqdm(trains, desc=origin) if verbose else trains):
if train.dest == destination:
# We hope that we have a direct train
valid_routes.append([train])
else:
if train.dest in explored and explored[train.dest][1] is None:
# This is a loop
continue
elif train.arr < min_dep:
# The train is not direct and arrives on the next day, we avoid that
continue
find_routes(day, train.dest, destination, verbose, train.arr, explored)
last_train = it[-1]
# Filter unusuable trains
valid_routes += [[train] + it for it in explored[train.dest][1] if it[0].dep >= train.arr]
if last_train.arr <= train.dep:
new_it = it + [train]
if train.dest == destination:
# Goal is achieved
valid_routes.append(new_it)
else:
explore.append(new_it)
per_arr_explore.setdefault(train.dest, [])
per_arr_explore[train.dest].append(new_it)
explored[origin] = (min_dep, valid_routes)
return valid_routes
# Send queued trains to the database
db.session.commit()
return {destination: valid_routes} if destination else {}
def print_route(route: list[Train]):
s = f"{route[0].orig} "
for tr in route:
s += f"({tr.dep}) --> ({tr.arr}) {tr.dest}, "
print(s[:-2])
# Don't use the decorator to keep the function callable
cli.command('find-routes')(click.argument('day', type=click.DateTime(formats=['%Y-%m-%d']))
(click.argument('origin', type=str)
(click.argument('destination', type=str, default=None)
(click.option('--verbose', '-v', type=bool, is_flag=True, help="Display errors.")
(find_routes)))))
@cli.command('queue-route')
@click.argument('day', type=click.DateTime(formats=['%Y-%m-%d']))
@click.argument('origin', type=str)
@click.argument('destination', type=str)
def queue_route(day: date | datetime, origin: str, destination: str):
def queue_route(day: date | datetime, origin: str, destination: str, verbose: bool = False, autocommit: bool = True):
"""
Fetch the TGVMax simulator to refresh data.
@ -221,44 +243,89 @@ def queue_route(day: date | datetime, origin: str, destination: str):
if isinstance(day, datetime):
day = day.date()
query = db.session.query(RouteQueue).filter_by(day=day, origin=origin, destination=destination, response_time=None)
query = db.session.query(RouteQueue).filter_by(day=day, origin=origin, destination=destination)\
.filter((RouteQueue.response_time == None) | (RouteQueue.expiration_time >= datetime.now(timezone('UTC'))))
if query.count():
print("Already queued")
return
query = db.session.query(RouteQueue).filter(RouteQueue.day == day,
RouteQueue.origin == origin,
RouteQueue.destination == destination,
RouteQueue.expiration_time >= datetime.now(timezone('UTC')))
if query.count():
print("Using recent value")
return
db.session.add(RouteQueue(day=day, origin=origin, destination=destination))
db.session.commit()
if autocommit:
db.session.commit()
# Don't use the decorator to keep the function callable
cli.command('queue-route')(click.argument('day', type=click.DateTime(formats=['%Y-%m-%d']))
(click.argument('origin', type=str)
(click.argument('destination', type=str)
(click.option('--verbose', '-v', type=bool, is_flag=True, help="Display errors.")
(queue_route)))))
def queue_routes(day: date | datetime, origin: str | None = None,
destination: str | None = None, verbose: bool = False,
autocommit: bool = True):
if isinstance(day, datetime):
day = day.date()
valid_routes = set(db.session.query(RouteQueue).filter_by(day=day)\
.filter((RouteQueue.response_time == None) | (RouteQueue.expiration_time >= datetime.now(timezone('UTC'))))\
.values(RouteQueue.origin, RouteQueue.destination))
query = db.session.query(Train).filter((Train.day == day))
if origin:
query = query.filter((Train.orig_iata == origin) | (Train.orig == origin))
if destination:
query = query.filter((Train.dest_iata == destination) | (Train.dest == destination))
query = query.all()
for train in (t := tqdm(query) if verbose else query):
if verbose:
t.set_description(f"{day}: {train.orig} --> {train.dest}")
if (train.orig_iata, train.dest_iata) not in valid_routes:
queue_route(day, train.orig_iata, train.dest_iata, verbose, autocommit)
valid_routes.add((train.orig_iata, train.dest_iata))
# Same as above
cli.command('queue-routes')(click.argument('day', type=click.DateTime(formats=['%Y-%m-%d']))
(click.option('--origin', '-o', default=None)
(click.option('--destination', '-d', default=None)
(click.option('--verbose', '-v', type=bool, is_flag=True, help="Display errors.")
(queue_routes)))))
@cli.command('process-queue', help="Process the waiting list to refresh from the simulator.")
@click.argument('number', default=5, type=int)
def process_queue(number: int):
queue = db.session.query(RouteQueue).filter_by(response_time=None).order_by(RouteQueue.queue_time)
@click.argument('number', default=30, type=int)
@click.option('--verbose', '-v', type=bool, is_flag=True, help="Display errors.")
def process_queue(number: int, verbose: bool = False):
queue = db.session.query(RouteQueue).filter_by(response_time=None).order_by(RouteQueue.queue_time).all()
if number > 0:
queue = queue[:number]
URL = "https://www.maxjeune-tgvinoui.sncf/api/public/refdata/search-freeplaces-proposals"
for req in queue:
if verbose:
query = db.session.query(Train).with_entities(Train.orig_iata, Train.orig).distinct()
iata_to_names = {k: v for (k, v) in query.all()}
for i, req in enumerate(t := tqdm(queue) if verbose else queue):
req: RouteQueue
resp = requests.post(URL, json={
'departureDateTime': req.day.isoformat(),
'origin': req.origin,
'destination': req.destination,
})
if verbose:
t.set_description(f"{req.day:%d/%m/%Y}: {iata_to_names[req.origin]} --> {iata_to_names[req.destination]}")
resp = None
while resp is None or resp.status_code == 429:
resp = requests.post(URL, json={
'departureDateTime': req.day.isoformat(),
'origin': req.origin,
'destination': req.destination,
})
if resp.status_code == 429:
sleep(1)
if resp.status_code == 404:
# No travel found
req.response_time = datetime.now()
req.expiration_time = datetime.now() + timedelta(hours=1)
req.expiration_time = datetime.now() + timedelta(hours=3)
db.session.add(req)
continue
@ -270,18 +337,27 @@ def process_queue(number: int):
req.expiration_time += timedelta(hours=3) # By default 5 minutes, extend it to 3 hours to be safe
db.session.add(req)
db.session.query(Train).filter_by(day=req.day, orig_iata=req.origin, dest_iata=req.destination)\
db.session.query(Train).filter_by(day=req.day, orig_iata=req.origin, dest_iata=req.destination) \
.update(dict(tgvmax=False, remaining_seats=-1))
for proposal in data['proposals']:
train = db.session.query(Train).filter_by(day=req.day, number=int(proposal['trainNumber']),
orig_iata=req.origin, dest_iata=req.destination).first()
orig_iata=proposal['origin']['rrCode'],
dest_iata=proposal['destination']['rrCode']).first()
if train is None:
# In a city with multiple stations
print("ERROR")
print(proposal)
continue
train.tgvmax = True
train.remaining_seats = proposal['freePlaces']
train.last_modification = req.response_time
train.expiration_time = req.expiration_time
db.session.add(train)
if i % 50 == 0:
db.session.commit()
db.session.commit()
@ -304,20 +380,29 @@ def iata_codes():
@app.get('/api/routes/<day>/<origin>/<destination>/')
def get_routes(day: date, origin: str, destination: str):
def get_routes(day: date | str, origin: str, destination: str):
if isinstance(day, str):
day = date.fromisoformat(day)
if destination == 'undefined':
destination = None
routes = find_routes(day, origin, destination)
return [
[{
'origin': tr.orig,
'origin_iata': tr.orig_iata,
'destination': tr.dest,
'destination_iata': tr.dest_iata,
'departure': tr.dep.isoformat(),
'arrival': tr.arr.isoformat(),
'number': tr.number,
'free_seats': tr.remaining_seats,
} for tr in route] for route in routes
]
return {
city: [
[{
'origin': tr.orig,
'origin_iata': tr.orig_iata,
'destination': tr.dest,
'destination_iata': tr.dest_iata,
'departure': tr.dep.isoformat(),
'arrival': tr.arr.isoformat(),
'number': tr.number,
'free_seats': tr.remaining_seats,
} for tr in route] for route in city_routes
] for city, city_routes in routes.items()
}
if __name__ == '__main__':

1
static/bootstrap5 Symbolic link
View File

@ -0,0 +1 @@
/usr/share/javascript/bootstrap5/

View File

@ -2,53 +2,50 @@
<html lang="fr">
<head>
<title>Calculateur TGVMax</title>
<link href="https://cdn.jsdelivr.net/npm/bootstrap@5.3.0-alpha1/dist/css/bootstrap.min.css"
rel="stylesheet"
integrity="sha384-GLhlTQ8iRABdZLl6O3oVMWSktQOp6b7In1Zl3/Jr59b6EGGoI1aFkw7cmDA6j6gD"
crossorigin="anonymous">
<link href="/static/bootstrap5/css/bootstrap.min.css" rel="stylesheet">
</head>
<body>
<form id="form" action="#">
<main class="container">
<form id="form" action="#">
<datalist id="iataCodes">
<option value="Chargement…">
</datalist>
<div class="mb-3">
<label for="origin" class="form-label">Origine :</label>
<input type="text" class="form-control" list="iataCodes" id="origin"
placeholder="Origine…" aria-describedby="originHelp">
placeholder="Origine…" aria-describedby="originHelp">
<input type="hidden" class="form-control" id="originIata">
<div id="originHelp" class="form-text">Le point de départ de votre trajet.</div>
</div>
<div class="mb-3">
<label for="destination" class="form-label">Destination :</label>
<input type="text" class="form-control" list="iataCodes" id="destination"
placeholder="Destination…" aria-describedby="destinationHelp">
placeholder="Destination…" aria-describedby="destinationHelp">
<input type="hidden" class="form-control" id="destinationIata">
<div id="destinationHelp" class="form-text">Le point d'arrivée de votre trajet.</div>
</div>
<div class="mb-3">
<label for="day" class="form-label">Jour de départ :</label>
<input type="date" class="form-control" id="day" aria-describedby="dayHelp"
min="{{ today }}" max="{{ max_day }}" value="{{ today }}">
min="{{ today }}" max="{{ max_day }}" value="{{ today }}">
<div id="dayHelp" class="form-text">Le jour de votre départ.</div>
</div>
<div class="mb-3">
<input type="submit" class="form-control btn btn-primary" value="Rechercher…">
</div>
</form>
<div id="result"></div>
</form>
<div id="result"></div>
</main>
</body>
<script src="https://cdn.jsdelivr.net/npm/bootstrap@5.3.0-alpha1/dist/js/bootstrap.bundle.min.js"
integrity="sha384-w76AqPfDkMBDXo30jS1Sgez6pr3x5MlQ1ZAGC+nuZB+EYdgRZgiwxhTBTkF7CXvN"
crossorigin="anonymous"></script>
<script src="/static/bootstrap5/js/bootstrap.bundle.min.js"></script>
<script>
fetch('/api/iata-codes/').then(res => res.json()).then(out => {
let datalist = document.getElementById('iataCodes')
datalist.innerHTML = ''
for (let iata in out.iata2name) {
let name = out.iata2name[iata]
for (let name in out.name2iata) {
let iata = out.name2iata[name]
let elem = document.createElement('option')
elem.value = name
elem.setAttribute('data-iata', iata)
@ -77,27 +74,71 @@
let day_elem = document.getElementById('day')
let result_elem = document.getElementById('result')
document.getElementById('form').addEventListener('submit', () => {
result_elem.innerHTML = 'Chargement…'
fetch('/api/routes/' + day_elem.value + '/' + origin_elem.value + '/' + destination_elem.value + '/')
result_elem.innerHTML = '<div class="spinner-border" role="status"><span class="visually-hidden">Chargement…</span></div>'
fetch('/api/routes/' + day_elem.value + '/' + origin_elem.value + '/' + (destination_elem.value || 'undefined') + '/')
.then(resp => resp.json())
.then(routes => {
console.log(routes)
result_elem.innerHTML = ''
let routes_elem = document.createElement('ul')
result_elem.appendChild(routes_elem)
for (let route in routes) {
route = routes[route]
console.log(route)
let route_elem = document.createElement('li')
routes_elem.appendChild(route_elem)
let text = route[0].origin
for (let train in route) {
train = route[train]
console.log(train)
text += " (" + train.departure + ") --> (" + train.arrival + ") " + train.destination + ", "
let cities_number = Object.keys(routes).length
let accordion = document.createElement('div')
accordion.classList.add('accordion')
accordion.setAttribute('id', 'result-accordion')
result_elem.appendChild(accordion)
for (let city in routes) {
let city_routes = routes[city]
let city_id = city.replaceAll(' ', '_')
let city_elem = document.createElement('div')
city_elem.classList.add('accordion-item')
accordion.appendChild(city_elem)
let city_header = document.createElement('h2')
city_header.classList.add('accordion-header')
city_header.setAttribute('id', 'collapse-header-' + city_id)
city_elem.appendChild(city_header)
let city_name_button = document.createElement('button')
city_name_button.classList.add('accordion-button')
if (cities_number > 1)
city_name_button.classList.add('collapsed')
city_name_button.setAttribute('type', 'button')
city_name_button.setAttribute('data-bs-toggle', 'collapse')
city_name_button.setAttribute('data-bs-target', '#collapse-' + city_id)
city_name_button.setAttribute('aria-expanded', cities_number > 1 ? 'false' : 'true')
city_name_button.setAttribute('aria-controls', 'collapse-' + city_id)
city_name_button.textContent = city
city_header.appendChild(city_name_button)
let collapse = document.createElement('div')
collapse.setAttribute('id', 'collapse-' + city_id)
collapse.classList.add('accordion-collapse')
if (cities_number > 1)
collapse.classList.add('collapse')
collapse.setAttribute('aria-labelled-by', 'collapse-header-' + city_id)
collapse.setAttribute('data-bs-parent', '#result-accordion')
city_elem.appendChild(collapse)
let collapse_body = document.createElement('div')
collapse_body.classList.add('accordion-body')
collapse.appendChild(collapse_body)
let routes_elem = document.createElement('ul')
collapse_body.appendChild(routes_elem)
for (let route of city_routes) {
let route_elem = document.createElement('li')
routes_elem.appendChild(route_elem)
let text = route[0].origin
for (let train of route) {
text += " (" + train.departure + ") --> (" + train.arrival + ") " + train.destination + " "
}
route_elem.textContent = text
}
route_elem.textContent = text
}
})
})