import datetime from io import BytesIO import pandas as pd from django.contrib.gis.measure import Distance from django.db.models import F from postamates.settings import DEFAULT_PLACEMENT_POINT_UPDATE_RADIUS, AGE_DAY_LIMIT from service import models from service.enums import PointStatus from service.tasks import raschet from service.utils import create_columns_dist, run_psql_command import base64 import requests from postamates.settings import GEOCODER_API_KEY from service.enums import MatchingStatus from django.contrib.gis.db.models.functions import Distance as Dist from django.db.models import Avg, Sum, Count class PointService: def update_fact(self, postamat_id: str, fact: int): qs = self.get_point_by_postamat_id(postamat_id) qs.update(**{'fact': fact}) def update_postamat_id(self, point_id: int, postamat_id: str): qs = self.get_point_by_id(point_id) qs.update(**{'postamat_id': postamat_id}) def start_mathing(self, obj_id: int): file = models.TempFiles.objects.get(id=obj_id) excel_file = base64.b64decode(file.data) df = pd.read_excel(excel_file) total = df.shape[0] matched = 0 problem = 0 for _i, row in df.iterrows(): addr = row['Адрес'] cat = row['Категория объекта'] req_url = f"https://geocode.search.hereapi.com/v1/geocode?q={addr}&apiKey={GEOCODER_API_KEY}" response = requests.get(req_url).json().get('items') if not response: models.PrePlacementPoint.objects.create(address=addr, matching_status=MatchingStatus.Error.name) problem += 1 continue coords = response[0]['position'] wkt = "POINT(" + str(coords['lng']) + " " + str(coords['lat']) + ")" response = response[0]['address'] obj = models.PlacementPoint.objects.filter(street=response['street'], house_number=response['houseNumber'], category=cat).values().first() if obj: obj.pop('id') models.PrePlacementPoint.objects.create(**{**obj, "matching_status": MatchingStatus.Matched.name}) matched += 1 else: models.PrePlacementPoint.objects.create(address=addr, street=response['street'], house_number=response['houseNumber'], category=cat, geometry=wkt, matching_status=MatchingStatus.New.name) models.TempFiles.objects.all().delete() return total, matched, problem @staticmethod def make_enrichment(): points = models.PrePlacementPoint.objects.filter(matching_status=MatchingStatus.New.name).all() groups = models.Post_and_pvzGroup.objects.all() for point in points: origin = point.geometry qs = models.PlacementPoint.objects.filter(status=PointStatus.Working.name).annotate( dist=Dist('geometry', origin)).order_by('dist') point.target_dist = qs[0].dist.m point.target_post_cnt = qs.filter( dist__lt=Distance(m=DEFAULT_PLACEMENT_POINT_UPDATE_RADIUS) ).count() point.target_cnt_ao_mean = qs[0].target_cnt_ao_mean point.rival_post_cnt = models.Post_and_pvz.objects.filter( category__name="Постамат", include_in_ml=True, wkt__distance_lt=(origin, Distance(m=DEFAULT_PLACEMENT_POINT_UPDATE_RADIUS))).count() point.rival_pvz_cnt = models.Post_and_pvz.objects.filter( category__name="ПВЗ", include_in_ml=True, wkt__distance_lt=(origin, Distance(m=DEFAULT_PLACEMENT_POINT_UPDATE_RADIUS))).count() point.metro_dist = models.OtherObjects.objects.filter(group__name='metro_stations').annotate( dist=Dist('wkt', origin)).order_by('dist')[0].dist.m point.property_price_bargains = models.OtherObjects.objects.filter( group__name="bargains", wkt__distance_lt=(origin, Distance(m=DEFAULT_PLACEMENT_POINT_UPDATE_RADIUS))).aggregate(Avg('param1'))[ 'param1__avg'] offers_estate = models.OtherObjects.objects.filter( group__name="offers_estate", wkt__distance_lt=(origin, Distance(m=DEFAULT_PLACEMENT_POINT_UPDATE_RADIUS))).aggregate( param1__avg=Avg('param1'), param3__avg=Avg('param3')) point.property_price_offers = offers_estate['param1__avg'] point.property_mean_floor = offers_estate['param3__avg'] point.property_era = models.OtherObjects.objects.filter( group__name="offers_estate").values('param2').annotate(cnt=Count('param2')).order_by('-cnt').first()[ 'param2'] point.flats_cnt = models.OtherObjects.objects.filter( group__name="flats_cnt", wkt__distance_lt=(origin, Distance(m=DEFAULT_PLACEMENT_POINT_UPDATE_RADIUS))).aggregate( param1__sum=Sum('param1'))['param1__sum'] popul_home_job = models.OtherObjects.objects.filter( group__name="popul_home_job", wkt__distance_lt=(origin, Distance(m=DEFAULT_PLACEMENT_POINT_UPDATE_RADIUS))).aggregate( param1__sum=Sum('param1'), param3__sum=Sum('param3')) point.popul_home = popul_home_job['param1__sum'] point.popul_job = popul_home_job['param3__sum'] yndx_food_cnt_amt = models.OtherObjects.objects.filter( group__name="yndx_food_cnt_amt", wkt__distance_lt=(origin, Distance(m=DEFAULT_PLACEMENT_POINT_UPDATE_RADIUS))).aggregate( param1__sum=Sum('param1'), param3__sum=Sum('param3')) point.yndxfood_sum = yndx_food_cnt_amt['param1__sum'] point.yndxfood_cnt = yndx_food_cnt_amt['param3__sum'] point.school_cnt = models.OtherObjects.objects.filter( group__name="schools", wkt__distance_lt=(origin, Distance(m=DEFAULT_PLACEMENT_POINT_UPDATE_RADIUS))).count() point.kindergar_cnt = models.OtherObjects.objects.filter( group__name="kindergar", wkt__distance_lt=(origin, Distance(m=DEFAULT_PLACEMENT_POINT_UPDATE_RADIUS))).count() point.public_stop_cnt = models.OtherObjects.objects.filter( group__name="stops", wkt__distance_lt=(origin, Distance(m=DEFAULT_PLACEMENT_POINT_UPDATE_RADIUS))).count() point.sport_center_cnt = models.OtherObjects.objects.filter( group__name="sport_centers", wkt__distance_lt=(origin, Distance(m=DEFAULT_PLACEMENT_POINT_UPDATE_RADIUS))).count() point.pharmacy_cnt = models.OtherObjects.objects.filter( group__name="pharmacies", wkt__distance_lt=(origin, Distance(m=DEFAULT_PLACEMENT_POINT_UPDATE_RADIUS))).count() point.supermarket_cnt = models.OtherObjects.objects.filter( group__name="supermarkets", wkt__distance_lt=(origin, Distance(m=DEFAULT_PLACEMENT_POINT_UPDATE_RADIUS))).count() point.supermarket_premium_cnt = models.OtherObjects.objects.filter( group__name="supermarkets_premium", wkt__distance_lt=(origin, Distance(m=DEFAULT_PLACEMENT_POINT_UPDATE_RADIUS))).count() point.clinic_cnt = models.OtherObjects.objects.filter( group__name="clinics", wkt__distance_lt=(origin, Distance(m=DEFAULT_PLACEMENT_POINT_UPDATE_RADIUS))).count() point.bank_cnt = models.OtherObjects.objects.filter( group__name="banks", wkt__distance_lt=(origin, Distance(m=DEFAULT_PLACEMENT_POINT_UPDATE_RADIUS))).count() point.reca_cnt = models.OtherObjects.objects.filter( group__name="recas", wkt__distance_lt=(origin, Distance(m=DEFAULT_PLACEMENT_POINT_UPDATE_RADIUS))).count() point.lab_cnt = models.OtherObjects.objects.filter( group__name="labs", wkt__distance_lt=(origin, Distance(m=DEFAULT_PLACEMENT_POINT_UPDATE_RADIUS))).count() point.culture_cnt = models.OtherObjects.objects.filter( group__name="cultures", wkt__distance_lt=(origin, Distance(m=DEFAULT_PLACEMENT_POINT_UPDATE_RADIUS))).count() point.attraction_cnt = models.OtherObjects.objects.filter( group__name="attractions", wkt__distance_lt=(origin, Distance(m=DEFAULT_PLACEMENT_POINT_UPDATE_RADIUS))).count() point.mfc_cnt = models.OtherObjects.objects.filter( group__name="public_services", wkt__distance_lt=(origin, Distance(m=DEFAULT_PLACEMENT_POINT_UPDATE_RADIUS))).count() point.bc_cnt = models.OtherObjects.objects.filter( group__name="BC", wkt__distance_lt=(origin, Distance(m=DEFAULT_PLACEMENT_POINT_UPDATE_RADIUS))).count() point.tc_cnt = models.OtherObjects.objects.filter( group__name="TC", wkt__distance_lt=(origin, Distance(m=DEFAULT_PLACEMENT_POINT_UPDATE_RADIUS))).count() point.business_activity = models.OtherObjects.objects.filter( group__name="business_activity", wkt__distance_lt=(origin, Distance(m=DEFAULT_PLACEMENT_POINT_UPDATE_RADIUS))).aggregate( param1__sum=Sum('param1'))['param1__sum'] point.age_day = AGE_DAY_LIMIT point.save() for group in groups: post_object = models.Post_and_pvz.objects.filter(group__name=group.name).annotate( distance=Dist("wkt", point.geometry)).order_by('distance').first() d = models.PrePlacementPointPVZDistance.objects.filter(placement_point=point, pvz_postamates_group=group).first() if post_object: if d: if d.dist > post_object.distance.m: d.dist = post_object.distance.m d.save() else: models.PrePlacementPointPVZDistance.objects.create(placement_point=point, pvz_postamates_group=group, dist=post_object.distance.m) run_psql_command() @staticmethod def get_min_distances_to_group(postamat_id: str): return {d['pvz_postamates_group']: d['dist'] for d in list( models.PlacementPointPVZDistance.objects.filter(placement_point=postamat_id).values( 'pvz_postamates_group', 'dist'))} @staticmethod def update_points_in_radius(qs: models.PlacementPoint, new_status: str): triggers = False for point in qs: if new_status == PointStatus.Installation.name: if point.status == PointStatus.Pending.name: pnts = models.PlacementPoint.objects.filter( geometry__distance_lt=(point.geometry, Distance(m=DEFAULT_PLACEMENT_POINT_UPDATE_RADIUS)), ) pnts.update(prediction_first=F('prediction_current'), target_post_cnt=F('target_post_cnt') + 1) triggers = True elif new_status == PointStatus.Cancelled.name or new_status == PointStatus.Pending.name: if point.status == PointStatus.Installation.name: pnts = models.PlacementPoint.objects.filter( geometry__distance_lt=(point.geometry, Distance(m=DEFAULT_PLACEMENT_POINT_UPDATE_RADIUS)), ) pnts.update(target_post_cnt=F('target_post_cnt') - 1 if F('target_post_cnt') != 0 else 0) triggers = True elif new_status == PointStatus.Working.name and point.status == PointStatus.Pending.name: triggers = True if triggers: raschet.delay() @staticmethod def update_status(qs: models.PlacementPoint, new_status: str) -> models.PlacementPoint: for q in qs: if q.status == PointStatus.Installation.name and new_status == PointStatus.Working.name: qs.update(**{'age_day': 0, 'start_date': datetime.datetime.now(), 'status': new_status}) else: qs.update(**{'status': new_status}) @staticmethod def get_point_by_id(point_id: int): return models.PlacementPoint.objects.filter(pk=point_id) @staticmethod def get_point_by_postamat_id(postamat_id: str): return models.PlacementPoint.objects.filter(postamat_id=postamat_id) @staticmethod def to_excel(serializer): data = pd.DataFrame(serializer.data) if not data.empty: if data['start_date'].any(): data['start_date'] = data.get('start_date').dt.tz_localize(None) if data['sample_trn'].any(): data['sample_trn'] = data['sample_trn'].astype(int) data.rename(columns={'district_id': 'district', 'area_id': 'area'}, inplace=True) data['min_distance_to_group'] = data['min_distance_to_group'].apply(lambda x: list(x.items())) new_columns = data.apply(create_columns_dist, axis=1) for ind in new_columns.columns: expanded = new_columns[ind].apply(pd.Series) group = models.Post_and_pvzGroup.objects.get(id=int(expanded.loc[0, 0])) expanded[[f"group_{ind + 1}_name", f"group_{ind + 1}_category"]] = group.name, group.category.name expanded = expanded.rename(columns={1: f"dist_to_group_{ind + 1}"}) expanded = expanded.drop(0, axis=1) new_columns = pd.concat([new_columns, expanded], axis=1) new_columns = new_columns.drop(ind, axis=1) data.drop('min_distance_to_group', axis=1, inplace=True) data = pd.concat([data, new_columns], axis=1) with BytesIO() as b: with pd.ExcelWriter(b) as writer: data.to_excel( writer, sheet_name='Placement Points', index=False, ) return b.getvalue() @staticmethod def to_json(serializer): data = pd.DataFrame(serializer.data) data['start_date'] = pd.to_datetime(data['start_date'], errors='coerce') data['start_date'] = data['start_date'].dt.tz_localize(None) data['sample_trn'] = data['sample_trn'].astype(int) data['geometry'] = data['geometry'].apply(lambda x: {'latitude': x[1], 'longtitude': x[0]}) data.rename(columns={'district_id': 'district', 'area_id': 'area'}, inplace=True) data['min_distance_to_group'] = data['min_distance_to_group'].apply(lambda x: list(x.items())) new_columns = data.apply(create_columns_dist, axis=1) for ind in new_columns.columns: expanded = new_columns[ind].apply(pd.Series) group = models.Post_and_pvzGroup.objects.get(id=int(expanded.loc[0, 0])) expanded[[f"group_{ind + 1}_name", f"group_{ind + 1}_category"]] = group.name, group.category.name expanded = expanded.rename(columns={1: f"dist_to_group_{ind + 1}"}) expanded = expanded.drop(0, axis=1) new_columns = pd.concat([new_columns, expanded], axis=1) new_columns = new_columns.drop(ind, axis=1) data.drop('min_distance_to_group', axis=1, inplace=True) data = pd.concat([data, new_columns], axis=1) return data.to_json(orient='records') @staticmethod def get_first_10_k(): if models.PlacementPoint.objects.count() > 10000: qs = models.PlacementPoint.objects.order_by('-prediction_current').all()[10000] return qs.prediction_current else: return models.PlacementPoint.objects.order_by('prediction_current').first().prediction_current