gault-millau/apps/location/transfer_data.py
2020-02-05 13:58:48 +03:00

812 lines
30 KiB
Python

import csv
import json
from pprint import pprint
from django.conf import settings
from django.db.transaction import atomic
from requests import get
from tqdm import tqdm
from account.models import Role
from collection.models import Collection
from gallery.models import Image
from location.models import Country, Region, City, Address
from main.models import AwardType
from news.models import News
from review.models import Review
from tag.models import TagCategory, ChosenTagSettings
from transfer import models as transfer_models
from transfer.serializers import location as location_serializers
from transfer.utils import clean_old_records, clean_old_country_records, clean_old_region_records
def transfer_countries():
queryset = transfer_models.Cities.objects.raw("""
SELECT cities.id, cities.country_code_2
FROM cities
WHERE country_code_2 IS NOT NULL AND
country_code_2 != ""
GROUP BY cities.id, cities.country_code_2
""")
queryset = [vars(query) for query in queryset]
serialized_data = location_serializers.CountrySerializer(data=queryset, many=True)
if serialized_data.is_valid():
serialized_data.save()
else:
pprint(f"Country serializer errors: {serialized_data.errors}")
def transfer_regions():
regions_without_subregion_queryset = transfer_models.Cities.objects.raw("""
SELECT cities.id,
cities.region_code,
cities.country_code_2,
cities.subregion_code
FROM cities
WHERE (subregion_code IS NULL
OR subregion_code = ""
)
AND region_code IS NOT NULL
AND region_code != ""
AND country_code_2 IS NOT NULL
AND country_code_2 != ""
GROUP BY cities.id,
cities.region_code,
cities.country_code_2,
cities.subregion_code
""")
regions_without_subregion_queryset = [vars(query) for query in regions_without_subregion_queryset]
serialized_without_subregion = location_serializers.RegionSerializer(data=regions_without_subregion_queryset,
many=True)
if serialized_without_subregion.is_valid():
serialized_without_subregion.save()
else:
pprint(f"Parent regions serializer errors: {serialized_without_subregion.errors}")
regions_with_subregion_queryset = transfer_models.Cities.objects.raw("""
SELECT
cities.id,
cities.region_code,
cities.country_code_2,
cities.subregion_code
FROM cities
WHERE subregion_code IS NOT NULL AND
subregion_code != "" AND
region_code IS NOT NULL AND
region_code != "" AND
country_code_2 IS NOT NULL AND
country_code_2 != ""
AND cities.subregion_code in
(
SELECT region_code FROM cities WHERE
(subregion_code IS NULL OR
subregion_code = "") AND
region_code IS NOT NULL AND
region_code != "" AND
country_code_2 IS NOT NULL AND
country_code_2 != ""
)
GROUP BY cities.id,
cities.region_code,
cities.country_code_2,
cities.subregion_code
""")
regions_with_subregion_queryset = [vars(query) for query in regions_with_subregion_queryset]
serialized_with_subregion = location_serializers.RegionSerializer(data=regions_with_subregion_queryset,
many=True)
if serialized_with_subregion.is_valid():
serialized_with_subregion.save()
else:
pprint(f"Child regions serializer errors: {serialized_with_subregion.errors}")
def transfer_cities():
queryset = transfer_models.Cities.objects.raw("""SELECT cities.id, cities.name, cities.country_code_2, cities.zip_code,
cities.is_island, cities.region_code, cities.subregion_code
FROM cities WHERE
region_code IS NOT NULL AND
region_code != "" AND
country_code_2 IS NOT NULL AND
country_code_2 != ""
""")
queryset = [vars(query) for query in queryset]
serialized_data = location_serializers.CitySerializer(data=queryset, many=True)
if serialized_data.is_valid():
serialized_data.save()
else:
pprint(f"City serializer errors: {serialized_data.errors}")
@atomic
def transfer_addresses():
queryset = transfer_models.Locations.objects.raw("""SELECT locations.id, locations.zip_code, locations.longitude,
locations.latitude, locations.address, locations.city_id
FROM locations WHERE
locations.city_id IS NOT NULL""")
queryset = [vars(query) for query in queryset]
serialized_data = location_serializers.AddressSerializer(data=queryset, many=True)
if serialized_data.is_valid():
serialized_data.save()
else:
pprint(f"Address serializer errors: {serialized_data.errors}")
def transfer_wine_region():
queryset = transfer_models.WineLocations.objects.filter(type='WineRegion')
serialized_data = location_serializers.WineRegionSerializer(
data=list(queryset.values()),
many=True)
if serialized_data.is_valid():
serialized_data.save()
else:
pprint(f"WineRegionSerializer errors: {serialized_data.errors}")
def transfer_wine_sub_region():
queryset = transfer_models.WineLocations.objects.filter(type='WineSubRegion')
serialized_data = location_serializers.WineSubRegionSerializer(
data=list(queryset.values()),
many=True)
if serialized_data.is_valid():
serialized_data.save()
else:
pprint(f"WineStandardClassificationSerializer errors: {serialized_data.errors}")
def transfer_wine_village():
queryset = transfer_models.WineLocations.objects.filter(type='Village')
serialized_data = location_serializers.WineVillageSerializer(
data=list(queryset.values()),
many=True)
if serialized_data.is_valid():
serialized_data.save()
else:
pprint(f"WineStandardClassificationSerializer errors: {serialized_data.errors}")
def update_flags():
queryset = Country.objects.only("id", "code", "svg_image").filter(old_id__isnull=False)
link_to_request = "https://s3.eu-central-1.amazonaws.com/gm-test.com/media"
for query in queryset:
svg_link = f"/svg/country/10-31-2019/{query.code}.svg"
resp = get(f"{link_to_request}{svg_link}")
if resp.status_code == 200:
query.svg_image = svg_link
query.save()
def update_fake_country_flag():
link_to_request = "https://s3.eu-central-1.amazonaws.com/gm-test.com/media/svg/country/10-31-2019/aa.svg"
resp = get(link_to_request)
if resp.status_code == 200:
country = Country.objects.get(code="aa")
country.svg_image = "/svg/country/10-31-2019/aa.svg"
country.save()
def migrate_city_map_situation(get_exists_cities=False):
if get_exists_cities:
ids = City.objects.values_list('mysql_id', flat=True)
queryset = transfer_models.Cities.objects.filter(id__in=list(ids))
queryset = list(queryset.values())
else:
queryset = transfer_models.Cities.objects.raw("""SELECT cities.id, cities.map1, cities.map2, cities.map_ref, cities.situation
FROM cities WHERE
region_code IS NOT NULL AND
region_code != "" AND
country_code_2 IS NOT NULL AND
country_code_2 != ""
""")
queryset = [vars(query) for query in queryset]
serialized_data = location_serializers.CityMapCorrectSerializer(data=queryset, many=True)
if serialized_data.is_valid():
serialized_data.save()
else:
pprint(f"City info serializer errors: {serialized_data.errors}")
# Update location models with ruby library
# Utils functions defined before transfer functions
def get_ruby_socket(params):
url = 'http://172.21.0.1:5678' # docker host
response = get(url, params=params)
try:
data = json.loads(response.text)
except Exception as e:
print(f"{response.text} error: {e}")
return None
return data
# Get data from ruby and save it to file
# Save errors from ruby to another file
def get_ruby_data():
cities = transfer_models.Cities.objects.all()
ruby_data = {}
error_file = open(f"{settings.PROJECT_ROOT}/ruby_error.txt", "w")
for mysql_city in cities:
# try:
# mysql_city = transfer_models.Cities.objects.get(id=city.old_id)
# except transfer_models.Cities.DoesNotExist:
# print(f"City with id {city.old_id} not found")
# continue
ruby_params = {}
if mysql_city.country_code:
ruby_params['country_code'] = mysql_city.country_code
if mysql_city.country_code_2:
ruby_params['country_code_2'] = mysql_city.country_code_2
if mysql_city.region_code:
ruby_params['region_code'] = mysql_city.region_code
if mysql_city.subregion_code:
ruby_params['subregion_code'] = mysql_city.subregion_code
ruby_response = get_ruby_socket(ruby_params) if ruby_params else None
if ruby_response is None:
continue
if "country" not in ruby_response.keys():
error_file.write(f"{json.dumps(ruby_params)}: {json.dumps(ruby_response)}\n")
continue
if mysql_city.country_code_2 is None and mysql_city.country_code is None:
error_file.write(f"{json.dumps(ruby_params)}: {json.dumps(ruby_response)}\n")
continue
city_name = mysql_city.name.strip()
ruby_data[mysql_city.id] = {
"mysql_id": mysql_city.id,
"name": {'en-GB': city_name.strip()},
"map1": mysql_city.map1,
"map2": mysql_city.map2,
"map_ref": mysql_city.map_ref,
"situation": mysql_city.situation,
"is_island": True if mysql_city.is_island is not None and mysql_city.is_island > 0 else False,
"postal_code": mysql_city.zip_code.strip() if mysql_city.zip_code is not None else '',
"code": ruby_response['country']['code'].lower().strip()
}
for ruby_data_key in ruby_response.keys():
if "error" in ruby_response[ruby_data_key]:
error_file.write(json.dumps(ruby_response) + '\n')
else:
ruby_data[mysql_city.id][ruby_data_key] = ruby_response[ruby_data_key]
error_file.close()
with open(f"{settings.PROJECT_ROOT}/apps/location/ruby_data.py", "w") as ruby_data_file:
ruby_data_file.write(json.dumps(ruby_data))
return ruby_data
def get_unused_data():
ruby_data = {}
error_file = open(f"{settings.PROJECT_ROOT}/ruby_unused_error.txt", "w")
countries = Country.objects.all()
for country in countries:
ruby_params = {}
ruby_response = get_ruby_socket({"country_regions": country.code})
if ruby_response is None:
continue
if "error" in ruby_response.keys():
error_file.write(f"{json.dumps(ruby_params)}: {json.dumps(ruby_response)}\n")
continue
ruby_data[country.code] = ruby_response
error_file.close()
with open(f"{settings.PROJECT_ROOT}/apps/location/ruby_unused_data.py", "w") as ruby_data_file:
ruby_data_file.write(json.dumps(ruby_data))
# Add correct objects of Country, Region and City with mysql_ids array (Country, Region) and mysql_id (City)
def add_correct_location_models(ruby_data):
for mysql_id, city_object in tqdm(ruby_data.items()):
country_data = city_object['country']
country_code = country_data['code'] # i.e. fr
try:
country = Country.objects.get(
code=country_code,
mysql_ids__isnull=False,
)
country.mysql_ids.append(mysql_id)
country.save()
except Country.DoesNotExist:
country_data['mysql_ids'] = [mysql_id]
country_data['calling_code'] = settings.COUNTRY_CALLING_CODES.get(country_code)
country = Country.objects.create(**country_data)
city_object['country'] = country
if "region" in city_object:
region_data = city_object['region']
region_data['country'] = country
try:
region = Region.objects.get(code=region_data['code'], mysql_ids__isnull=False)
region.mysql_ids.append(mysql_id)
region.save()
except Region.DoesNotExist:
region_data['mysql_ids'] = [mysql_id]
region = Region.objects.create(**region_data)
if "subregion" in city_object:
subregion_data = city_object.pop('subregion')
subregion_data['country'] = country
try:
subregion = Region.objects.get(code=subregion_data['code'],
mysql_ids__isnull=False)
subregion.mysql_ids.append(mysql_id)
subregion.save()
except Region.DoesNotExist:
subregion_data['parent_region'] = region
subregion_data['mysql_ids'] = [mysql_id]
subregion = Region.objects.create(**subregion_data)
city_object['region'] = subregion
else:
city_object['region'] = region
if "subregion" in city_object:
del (city_object['subregion'])
try:
City.objects.create(**city_object)
except Exception as e:
print(city_object)
print(e)
break
def fix_location_address():
addresses = Address.objects.filter(old_id__isnull=False)
for address in addresses:
mysql_location = transfer_models.Locations.objects.get(id=address.old_id)
try:
correct_city = City.objects.get(mysql_id=mysql_location.city_id)
except City.DoesNotExist:
continue
address.city = correct_city
address.save()
def fix_location_collection():
collections = Collection.objects.filter(old_id__isnull=False)
for collection in collections:
try:
mysql_collection = transfer_models.Collections.objects. \
raw(f"""select
s.country_code_2,
c.id
from collections as c
join sites s on s.id = c.site_id where c.id={collection.old_id}""")[0]
except:
continue
try:
correct_country = Country.objects.get(code=mysql_collection.country_code_2, mysql_ids__isnull=False)
except Country.DoesNotExist:
continue
collection.country = correct_country
collection.save()
def fix_award_type():
award_types = AwardType.objects.filter(old_id__isnull=False)
for award_type in award_types:
mysql_award_type = transfer_models.AwardTypes.objects. \
raw(f"""SELECT at.id, s.country_code_2 AS country_code
FROM award_types as at
JOIN sites s on s.id = at.site_id
WHERE at.id={award_type.old_id}""")[0]
try:
correct_country = Country.objects.get(code=mysql_award_type.country_code, mysql_ids__isnull=False)
except Country.DoesNotExist:
continue
award_type.country = correct_country
award_type.save()
def fix_role():
roles_list = Role.objects.filter(country__isnull=False)
for role in roles_list:
try:
correct_country = Country.objects.get(code=role.country.code, mysql_ids__isnull=False)
except Country.DoesNotExist:
continue
role.country = correct_country
role.save()
def fix_news():
news = News.objects.filter(country__isnull=False)
for news_item in news:
try:
correct_country = Country.objects.get(code=news_item.country.code, mysql_ids__isnull=False)
except Country.DoesNotExist:
continue
news.country = correct_country
news_item.save()
def fix_reviews():
reviews = Review.objects.filter(country__isnull=False)
for review in reviews:
try:
correct_country = Country.objects.get(code=review.country.code, mysql_ids__isnull=False)
except Country.DoesNotExist:
continue
review.country = correct_country
review.save()
def fix_tag_category():
tags_categories = TagCategory.objects.filter(country__isnull=False)
for tag_category in tags_categories:
try:
correct_country = Country.objects.get(code=tag_category.country.code, mysql_ids__isnull=False)
except Country.DoesNotExist:
continue
tag_category.country = correct_country
tag_category.save()
def fix_chosen_tag():
chosen_tags = ChosenTagSettings.objects.filter(country__isnull=False)
for chosen_tag in chosen_tags:
try:
correct_country = Country.objects.get(code=chosen_tag.country.code, mysql_ids__isnull=False)
except Country.DoesNotExist:
continue
chosen_tag.country = correct_country
chosen_tag.save()
def fix_location_models():
try:
ruby_data_file = open(f"{settings.PROJECT_ROOT}/apps/location/ruby_data.py", "r")
ruby_data = json.loads(ruby_data_file.read())
except FileNotFoundError:
print('create ruby data')
ruby_data = get_ruby_data()
print('add_correct_location_models')
add_correct_location_models(ruby_data)
# print('fix_location_address')
# fix_location_address()
# print('fix_location_collection')
# fix_location_collection()
# print('fix_award_type')
# fix_award_type()
# print('fix_role')
# fix_role()
# print('fix_news')
# fix_news()
# print('fix_reviews')
# fix_reviews()
# print('fix_tag_category')
# fix_tag_category()
# print('fix_chosen_tag')
# fix_chosen_tag()
def remove_old_records():
clean_old_records(City, {"mysql_id__isnull": True})
clean_old_country_records(Country, {"mysql_ids__isnull": True})
clean_old_region_records(Region, {"mysql_ids__isnull": True})
def transfer_city_photos():
created_counter = 0
cities_not_exists = {}
cities_has_same_image = 0
city_gallery = transfer_models.CityPhotos.objects.exclude(city__isnull=True) \
.exclude(city__country_code_2__isnull=True) \
.exclude(city__country_code_2__iexact='') \
.exclude(city__region_code__isnull=True) \
.exclude(city__region_code__iexact='') \
.values_list('city_id', 'attachment_suffix_url')
for old_city_id, image_suffix_url in tqdm(city_gallery):
city = City.objects.filter(old_id=old_city_id)
if city.exists():
city = city.first()
image, _ = Image.objects.get_or_create(image=image_suffix_url,
defaults={
'image': image_suffix_url,
'orientation': Image.HORIZONTAL,
'title': f'{city.name} - {image_suffix_url}',
})
if city.image != image:
city.image = image
city.save()
created_counter += 1
else:
cities_has_same_image += 1
else:
cities_not_exists.update({'city_old_id': old_city_id})
print(f'Created: {created_counter}\n'
f'City not exists: {cities_not_exists}\n'
f'City has same image: {cities_has_same_image}')
@atomic
def add_fake_country():
# add country
country_data = {
"name": '{"en-GB": "Antilles Guyane West Indies"}',
"code": "aa",
"calling_code": "590",
"svg_image": "svg/country/11-02-2019/658714.svg",
"mysql_ids": []
}
country, _ = Country.objects.get_or_create(**country_data)
# add regions and subregions
regions_data = [
{
"name": "Guadeloupe",
"code": "gp",
"subregions": [
{
"name": "Basse Terre",
"code": "bat",
},
{
"name": "Grande Terre",
"code": "grt",
},
{
"name": "Autres les Îles de Guadeloupe",
"code": "aug",
},
]
},
{
"name": "Martinique",
"code": "mq",
"subregions": [
{
"name": "Nord",
"code": "nor",
},
{
"name": "Sud",
"code": "sud",
},
]
},
{
"name": "Guyane",
"code": "gy",
},
{
"name": "St Barthelemy",
"code": "bl",
},
{
"name": "St Martin",
"code": "mf",
},
{
"name": "Sainte-Lucie",
"code": "lc",
},
]
regions = {}
for region_data in regions_data:
if "subregions" in region_data:
subregions = region_data['subregions']
del (region_data['subregions'])
else:
subregions = False
region_name = region_data['name']
region_data['name'] = '{"en-GB": "' + region_name + '"}'
region_data['country'] = country
region_data['mysql_ids'] = []
regions[region_name] = Region.objects.create(**region_data)
if subregions:
for subregion_data in subregions:
subregion_name = subregion_data['name']
subregion_data['name'] = '{"en-GB": "' + subregion_name + '"}'
subregion_data['country'] = country
subregion_data['parent_region'] = regions[region_name]
subregion_data['mysql_ids'] = []
regions[subregion_name] = Region.objects.create(**subregion_data)
# add cities
file = open(f"{settings.PROJECT_ROOT}/apps/location/csv/aa_cities.csv")
reader = csv.DictReader(file, delimiter=',')
for city_data in reader:
del (city_data[''])
city_data['mysql_id'] = city_data['old_id']
del (city_data['old_id'])
city_data['postal_code'] = city_data['zip_code']
del (city_data['zip_code'])
if city_data['postal_code'] == 'null' or city_data['postal_code'] == '':
del (city_data['postal_code'])
city_data["country"] = country
if city_data['subregion'] != 'null':
region = regions[city_data['subregion']]
elif city_data['region'] != 'null':
region = regions[city_data['region']]
else:
del (city_data['region'])
region = None
del (city_data['subregion'])
city_data['region'] = region
city_data['name'] = '{"en-GB": "' + city_data['name'] + '"}'
city_data['is_island'] = True if int(city_data['is_island']) > 0 else False
if region is not None:
region.mysql_ids.append(city_data['mysql_id'])
country.mysql_ids.append(city_data['mysql_id'])
try:
mysql_data = transfer_models.Cities.objects. \
only("map1", "map2", "map_ref", "situation").get(id=city_data['mysql_id'])
city_data['map1'] = mysql_data.map1
city_data['map2'] = mysql_data.map2
city_data['map_ref'] = mysql_data.map_ref
city_data['situation'] = mysql_data.situation
except transfer_models.Cities.DoesNotExist:
pass
City.objects.create(**city_data)
country.save()
for region_name, region in regions.items():
if len(region.mysql_ids) > 0:
region.save()
@atomic
def setup_clean_db():
try:
ruby_data_file = open(f"{settings.PROJECT_ROOT}/apps/location/ruby_data.py", "r")
ruby_data = json.loads(ruby_data_file.read())
except FileNotFoundError:
ruby_data = get_ruby_data()
print('add_correct_location_models')
add_correct_location_models(ruby_data)
print('add_fake_country')
add_fake_country()
print('migrate_city_map_situation')
migrate_city_map_situation(True)
print('update_flags')
update_flags()
print('transfer_city_photos')
transfer_city_photos()
def set_unused_regions():
ruby_data_file = open(f"{settings.PROJECT_ROOT}/apps/location/ruby_unused_data.py", "r")
ruby_data = json.loads(ruby_data_file.read())
for country_code, regions in ruby_data.items():
try:
country = Country.objects.get(code=country_code)
except Country.DoesNotExist:
print(f"Country with code {country_code} does not exists")
continue
for region_code, region_obj in regions.items():
try:
region = Region.objects.get(code=region_code, country=country)
except Region.DoesNotExist:
region = Region.objects.create(
name=region_obj['name'],
code=region_code,
country=country
)
if "subregions" in region_obj:
for subregion_code, subregion in region_obj['subregions'].items():
try:
subregion = Region.objects.get(code=subregion, country=country)
except Region.DoesNotExist:
subregion = Region.objects.create(
name=subregion,
code=subregion_code,
country=country,
parent_region=region
)
data_types = {
"dictionaries": [
# transfer_countries, # !!! см setup_clean_db
# transfer_regions, # !!! см setup_clean_db
# transfer_cities, # !!! см setup_clean_db
transfer_addresses,
transfer_wine_region,
transfer_wine_sub_region,
transfer_wine_village,
],
"update_country_flag": [
update_flags
],
"update_city_info": [
migrate_city_map_situation
],
"fix_location": [
add_fake_country,
fix_location_models,
],
"remove_old_locations": [
remove_old_records
],
"migrate_city_photos": [
transfer_city_photos,
],
"add_fake_country": [
add_fake_country,
],
"setup_clean_db": [setup_clean_db],
"set_unused_regions": [set_unused_regions],
"update_fake_country_flag": [update_fake_country_flag]
}