ddadmin
9f69a33c31
- Зачем: - Superset dashboard открывался с ошибками datasources и неудобным layout, а курс не содержал готового урока по BI-витрине. - Что: - добавлен урок 6 про Superset поверх ClickHouse DM-витрин. - исправлена раскладка dashboard и дефолтный фильтр даты для исторических демо-данных. - добавлено восстановление metadata колонок датасетов при обновлении dashboard. - Проверка: - make superset-dashboard. - /api/v1/dashboard/1/datasets и /superset/explore_json для chart 10 возвращают 200. - python3 -m py_compile superset/create_dashboard.py; git diff --cached --check.
571 lines
22 KiB
Python
571 lines
22 KiB
Python
#!/usr/bin/env python3
|
||
"""
|
||
================================================================================
|
||
Скрипт создания дашборда "E-commerce Analytics" в Superset
|
||
================================================================================
|
||
Назначение:
|
||
- Создание чартов (Charts) на основе датасетов DM-слоя
|
||
- Создание дашборда с layout и native filters
|
||
|
||
Запуск:
|
||
Внутри контейнера superset:
|
||
python /app/superset_init/create_dashboard.py
|
||
================================================================================
|
||
"""
|
||
|
||
import os
|
||
import sys
|
||
import json
|
||
import logging
|
||
from datetime import datetime
|
||
|
||
logging.basicConfig(level=logging.INFO, format='%(asctime)s - %(levelname)s - %(message)s')
|
||
logger = logging.getLogger(__name__)
|
||
|
||
sys.path.insert(0, '/app')
|
||
|
||
|
||
# Конфигурация чартов
|
||
CHARTS_CONFIG = [
|
||
# KPI блок
|
||
{
|
||
"slice_name": "📊 Total Events",
|
||
"viz_type": "big_number",
|
||
"dataset_name": "v_events_enriched",
|
||
"params": {
|
||
"metric": {
|
||
"expressionType": "SQL",
|
||
"sqlExpression": "COUNT(*)",
|
||
"column": None,
|
||
"aggregate": None,
|
||
"label": "Total Events",
|
||
"optionName": "metric_1"
|
||
},
|
||
"granularity_sqla": "event_ts",
|
||
"y_axis_format": ",d",
|
||
"show_trend_line": False,
|
||
"time_range": "No filter"
|
||
}
|
||
},
|
||
{
|
||
"slice_name": "👤 Unique Users",
|
||
"viz_type": "big_number",
|
||
"dataset_name": "v_events_enriched",
|
||
"params": {
|
||
"metric": {
|
||
"expressionType": "SQL",
|
||
"sqlExpression": "COUNT(DISTINCT user_domain_id)",
|
||
"label": "Unique Users",
|
||
"optionName": "metric_2"
|
||
},
|
||
"granularity_sqla": "event_ts",
|
||
"y_axis_format": ",d",
|
||
"show_trend_line": False,
|
||
"time_range": "No filter"
|
||
}
|
||
},
|
||
{
|
||
"slice_name": "🎯 Unique Sessions",
|
||
"viz_type": "big_number",
|
||
"dataset_name": "v_events_enriched",
|
||
"params": {
|
||
"metric": {
|
||
"expressionType": "SQL",
|
||
"sqlExpression": "COUNT(DISTINCT click_id)",
|
||
"label": "Unique Sessions",
|
||
"optionName": "metric_3"
|
||
},
|
||
"granularity_sqla": "event_ts",
|
||
"y_axis_format": ",d",
|
||
"show_trend_line": False,
|
||
"time_range": "No filter"
|
||
}
|
||
},
|
||
{
|
||
"slice_name": "📈 Avg Events/Session",
|
||
"viz_type": "big_number",
|
||
"dataset_name": "v_events_enriched",
|
||
"params": {
|
||
"metric": {
|
||
"expressionType": "SQL",
|
||
"sqlExpression": "COUNT(*) / COUNT(DISTINCT click_id)",
|
||
"label": "Avg Events/Session",
|
||
"optionName": "metric_4"
|
||
},
|
||
"granularity_sqla": "event_ts",
|
||
"y_axis_format": ".2f",
|
||
"show_trend_line": False,
|
||
"time_range": "No filter"
|
||
}
|
||
},
|
||
# Динамика
|
||
{
|
||
"slice_name": "📅 Events by Hour",
|
||
"viz_type": "echarts_timeseries_line",
|
||
"dataset_name": "v_events_enriched",
|
||
"params": {
|
||
"granularity_sqla": "event_ts",
|
||
"time_grain_sqla": "PT1H",
|
||
"metrics": [
|
||
{
|
||
"expressionType": "SQL",
|
||
"sqlExpression": "COUNT(*)",
|
||
"label": "Events"
|
||
}
|
||
],
|
||
"groupby": [],
|
||
"time_range": "No filter",
|
||
"adhoc_filters": [],
|
||
"row_limit": 10000
|
||
}
|
||
},
|
||
{
|
||
"slice_name": "📱 Traffic by Device",
|
||
"viz_type": "pie",
|
||
"dataset_name": "v_events_enriched",
|
||
"params": {
|
||
"groupby": ["device_type"],
|
||
"metric": {
|
||
"expressionType": "SQL",
|
||
"sqlExpression": "COUNT(*)",
|
||
"label": "Count"
|
||
},
|
||
"row_limit": 100,
|
||
"donut": True,
|
||
"show_legend": True,
|
||
"labels_outside": True,
|
||
"time_range": "No filter"
|
||
}
|
||
},
|
||
# География
|
||
{
|
||
"slice_name": "🌍 Geography Map",
|
||
"viz_type": "world_map",
|
||
"dataset_name": "v_events_enriched",
|
||
"params": {
|
||
"entity": "geo_country",
|
||
"metric": {
|
||
"expressionType": "SQL",
|
||
"sqlExpression": "COUNT(*)",
|
||
"label": "Events"
|
||
},
|
||
"row_limit": 500,
|
||
"linear_color_scheme": "blue_white_yellow",
|
||
"time_range": "No filter"
|
||
}
|
||
},
|
||
# Маркетинг
|
||
{
|
||
"slice_name": "🔗 UTM Effectiveness Table",
|
||
"viz_type": "table",
|
||
"dataset_name": "v_utm_effectiveness",
|
||
"params": {
|
||
"groupby": ["utm_source", "utm_medium", "utm_campaign"],
|
||
"metrics": [
|
||
{"expressionType": "SQL", "sqlExpression": "SUM(clicks)", "label": "Clicks"},
|
||
{"expressionType": "SQL", "sqlExpression": "SUM(uniq_users)", "label": "Users"},
|
||
{"expressionType": "SQL", "sqlExpression": "SUM(uniq_sessions)", "label": "Sessions"}
|
||
],
|
||
"row_limit": 100,
|
||
"time_range": "No filter",
|
||
"adhoc_filters": [
|
||
{
|
||
"clause": "WHERE",
|
||
"expressionType": "SQL",
|
||
"sqlExpression": "utm_source IS NOT NULL",
|
||
"subject": None,
|
||
"operator": None,
|
||
"comparator": None
|
||
}
|
||
]
|
||
}
|
||
},
|
||
{
|
||
"slice_name": "📄 Top Pages",
|
||
"viz_type": "dist_bar",
|
||
"dataset_name": "v_top_pages_daily",
|
||
"params": {
|
||
"groupby": ["page_url_path"],
|
||
"metrics": [
|
||
{"expressionType": "SQL", "sqlExpression": "SUM(pageviews)", "label": "Pageviews"}
|
||
],
|
||
"row_limit": 20,
|
||
"time_range": "No filter",
|
||
"orientation": "vertical",
|
||
"show_legend": False
|
||
}
|
||
},
|
||
# Качество данных
|
||
{
|
||
"slice_name": "🔍 Data Quality Summary",
|
||
"viz_type": "dist_bar",
|
||
"dataset_name": "dq_summary",
|
||
"params": {
|
||
"groupby": ["layer"],
|
||
"metrics": [
|
||
{"expressionType": "SQL", "sqlExpression": "SUM(check_value)", "label": "Row Count"}
|
||
],
|
||
"adhoc_filters": [
|
||
{
|
||
"clause": "WHERE",
|
||
"expressionType": "SQL",
|
||
"sqlExpression": "check_name = 'total_rows'",
|
||
"subject": None,
|
||
"operator": None,
|
||
"comparator": None
|
||
}
|
||
],
|
||
"row_limit": 100,
|
||
"time_range": "No filter",
|
||
"show_legend": False
|
||
}
|
||
}
|
||
]
|
||
|
||
# Конфигурация дашборда
|
||
DASHBOARD_CONFIG = {
|
||
"dashboard_title": "🛒 E-commerce Analytics Dashboard",
|
||
"description": "Аналитический дашборд для e-commerce кликстрима: трафик, конверсии, география и качество данных.",
|
||
"published": True,
|
||
"slug": "ecommerce-analytics",
|
||
}
|
||
|
||
|
||
CHART_LAYOUT_BY_TITLE = {
|
||
"📊 Total Events": {"width": 3, "height": 32, "x": 0, "y": 0},
|
||
"👤 Unique Users": {"width": 3, "height": 32, "x": 3, "y": 0},
|
||
"🎯 Unique Sessions": {"width": 3, "height": 32, "x": 6, "y": 0},
|
||
"📈 Avg Events/Session": {"width": 3, "height": 32, "x": 9, "y": 0},
|
||
"📅 Events by Hour": {"width": 8, "height": 60, "x": 0, "y": 32},
|
||
"📱 Traffic by Device": {"width": 4, "height": 60, "x": 8, "y": 32},
|
||
"🌍 Geography Map": {"width": 6, "height": 60, "x": 0, "y": 92},
|
||
"🔗 UTM Effectiveness Table": {"width": 6, "height": 60, "x": 6, "y": 92},
|
||
"📄 Top Pages": {"width": 6, "height": 60, "x": 0, "y": 152},
|
||
"🔍 Data Quality Summary": {"width": 6, "height": 60, "x": 6, "y": 152},
|
||
}
|
||
|
||
|
||
def sync_query_context(chart, params: dict, dataset_id: int) -> None:
|
||
"""
|
||
Синхронизирует сохраненный query_context с обновленными params чарта.
|
||
|
||
Для Superset 4.x у `big_number` запрос валидируется как time-series и
|
||
ожидает `granularity` в query_context (проверено по актуальной документации).
|
||
"""
|
||
if not chart.query_context:
|
||
return
|
||
|
||
try:
|
||
query_context = json.loads(chart.query_context)
|
||
except (TypeError, json.JSONDecodeError):
|
||
logger.warning("Chart ID %s has invalid query_context, skip sync", chart.id)
|
||
return
|
||
|
||
query_context["datasource"] = {"id": dataset_id, "type": "table"}
|
||
query_context["form_data"] = {
|
||
**params,
|
||
"datasource": f"{dataset_id}__table",
|
||
"viz_type": chart.viz_type,
|
||
"slice_id": chart.id,
|
||
}
|
||
|
||
queries = query_context.get("queries")
|
||
if not isinstance(queries, list) or not queries:
|
||
chart.query_context = json.dumps(query_context)
|
||
return
|
||
|
||
if chart.viz_type == "big_number":
|
||
query = queries[0]
|
||
granularity = params.get("granularity_sqla")
|
||
if granularity:
|
||
query["granularity"] = granularity
|
||
query["is_timeseries"] = True
|
||
query["time_range"] = params.get("time_range", query.get("time_range"))
|
||
if "metric" in params:
|
||
query["metrics"] = [params["metric"]]
|
||
extras = query.get("extras") if isinstance(query.get("extras"), dict) else {}
|
||
if "time_grain_sqla" in params:
|
||
extras["time_grain_sqla"] = params.get("time_grain_sqla")
|
||
query["extras"] = extras
|
||
elif params.get("x_axis") or params.get("groupby"):
|
||
# Для категориальных графиков синхронизируем колонки измерений.
|
||
dimensions = params.get("groupby")
|
||
if not dimensions and params.get("x_axis"):
|
||
dimensions = [params["x_axis"]]
|
||
query = queries[0]
|
||
query["columns"] = dimensions
|
||
if "metrics" in params:
|
||
query["metrics"] = params["metrics"]
|
||
elif "metric" in params:
|
||
query["metrics"] = [params["metric"]]
|
||
query["row_limit"] = params.get("row_limit", query.get("row_limit"))
|
||
query["time_range"] = params.get("time_range", query.get("time_range"))
|
||
query["is_timeseries"] = False
|
||
|
||
chart.query_context = json.dumps(query_context)
|
||
|
||
|
||
def build_dashboard_metadata(filter_dataset_id: int | None) -> str:
|
||
"""Формирует json_metadata с валидными datasetId для native filters."""
|
||
native_filters = []
|
||
|
||
if filter_dataset_id is not None:
|
||
native_filters = [
|
||
{
|
||
"id": "date_filter",
|
||
"name": "📅 Date Range",
|
||
"filterType": "filter_time",
|
||
"targets": [{"datasetId": filter_dataset_id, "column": {"name": "event_date"}}],
|
||
"defaultValue": "No filter",
|
||
"scope": {"rootPath": ["ROOT_ID"], "excluded": []},
|
||
"cascadeParentIds": [],
|
||
"isInstant": True
|
||
},
|
||
{
|
||
"id": "country_filter",
|
||
"name": "🌍 Country",
|
||
"filterType": "filter_select",
|
||
"targets": [{"datasetId": filter_dataset_id, "column": {"name": "geo_country"}}],
|
||
"scope": {"rootPath": ["ROOT_ID"], "excluded": []},
|
||
"isInstant": True,
|
||
"allowsMultipleValues": True,
|
||
"isRequired": False
|
||
},
|
||
{
|
||
"id": "device_filter",
|
||
"name": "📱 Device Type",
|
||
"filterType": "filter_select",
|
||
"targets": [{"datasetId": filter_dataset_id, "column": {"name": "device_type"}}],
|
||
"scope": {"rootPath": ["ROOT_ID"], "excluded": []},
|
||
"isInstant": True,
|
||
"allowsMultipleValues": True,
|
||
"isRequired": False
|
||
},
|
||
{
|
||
"id": "browser_filter",
|
||
"name": "🌐 Browser",
|
||
"filterType": "filter_select",
|
||
"targets": [{"datasetId": filter_dataset_id, "column": {"name": "browser_name"}}],
|
||
"scope": {"rootPath": ["ROOT_ID"], "excluded": []},
|
||
"isInstant": True,
|
||
"allowsMultipleValues": True,
|
||
"isRequired": False
|
||
}
|
||
]
|
||
|
||
metadata = {
|
||
"native_filter_configuration": native_filters,
|
||
"color_scheme": "supersetColors",
|
||
"label_colors": {}
|
||
}
|
||
return json.dumps(metadata)
|
||
|
||
|
||
def main() -> bool:
|
||
"""Главная функция"""
|
||
logger.info("=" * 60)
|
||
logger.info("Creating E-commerce Analytics Dashboard")
|
||
logger.info("=" * 60)
|
||
|
||
# Импорты внутри main после создания app context
|
||
from superset.app import create_app
|
||
|
||
app = create_app()
|
||
|
||
with app.app_context():
|
||
from superset.extensions import db
|
||
from superset.models.slice import Slice
|
||
from superset.models.dashboard import Dashboard
|
||
from superset.connectors.sqla.models import SqlaTable
|
||
|
||
created_charts = []
|
||
datasets_by_name = {}
|
||
|
||
# Создаём чарты
|
||
for chart_config in CHARTS_CONFIG:
|
||
dataset = db.session.query(SqlaTable).filter_by(
|
||
table_name=chart_config["dataset_name"],
|
||
schema="dm"
|
||
).first()
|
||
|
||
if not dataset:
|
||
logger.warning(f"Dataset '{chart_config['dataset_name']}' not found, skipping chart")
|
||
continue
|
||
|
||
if not dataset.columns:
|
||
# Если dataset создали до DDL/DM, в metadata Superset нет колонок.
|
||
# Обновляем их здесь, чтобы dashboard восстанавливался через make superset-dashboard.
|
||
dataset.fetch_metadata()
|
||
db.session.commit()
|
||
logger.info("Refreshed dataset metadata: %s", dataset.table_name)
|
||
|
||
datasets_by_name[chart_config["dataset_name"]] = dataset.id
|
||
|
||
try:
|
||
# Подготавливаем параметры
|
||
params = chart_config["params"].copy()
|
||
params["datasource"] = f"{dataset.id}__table"
|
||
params["viz_type"] = chart_config["viz_type"]
|
||
serialized_params = json.dumps(params)
|
||
|
||
# Проверяем, существует ли уже чарт
|
||
existing = db.session.query(Slice).filter_by(
|
||
slice_name=chart_config["slice_name"]
|
||
).first()
|
||
|
||
if existing:
|
||
# Синхронизируем параметры существующего чарта с конфигом.
|
||
existing.viz_type = chart_config["viz_type"]
|
||
existing.datasource_id = dataset.id
|
||
existing.datasource_type = "table"
|
||
existing.datasource_name = dataset.table_name
|
||
existing.params = serialized_params
|
||
sync_query_context(existing, params, dataset.id)
|
||
existing.description = f"Chart created automatically for {chart_config['dataset_name']}"
|
||
db.session.flush()
|
||
logger.info(
|
||
f"Chart '{chart_config['slice_name']}' already exists (ID: {existing.id}), "
|
||
"params synced"
|
||
)
|
||
created_charts.append({"id": existing.id, "title": existing.slice_name})
|
||
continue
|
||
|
||
# Создаём чарт
|
||
chart = Slice(
|
||
slice_name=chart_config["slice_name"],
|
||
viz_type=chart_config["viz_type"],
|
||
datasource_id=dataset.id,
|
||
datasource_type="table",
|
||
datasource_name=dataset.table_name,
|
||
params=serialized_params,
|
||
description=f"Chart created automatically for {chart_config['dataset_name']}"
|
||
)
|
||
|
||
db.session.add(chart)
|
||
db.session.flush()
|
||
|
||
logger.info(f"Created chart: {chart_config['slice_name']} (ID: {chart.id})")
|
||
created_charts.append({"id": chart.id, "title": chart.slice_name})
|
||
|
||
except Exception as e:
|
||
logger.error(f"Failed to create chart '{chart_config['slice_name']}': {e}")
|
||
import traceback
|
||
traceback.print_exc()
|
||
db.session.rollback()
|
||
|
||
logger.info(f"Created/Found {len(created_charts)} charts")
|
||
metadata_json = build_dashboard_metadata(datasets_by_name.get("v_events_enriched"))
|
||
|
||
# Создаём позиции чартов для layout.
|
||
# Обязательные блоки ROOT_ID/GRID_ID нужны для корректной работы /tabs.
|
||
positions = {
|
||
"DASHBOARD_VERSION_KEY": "v2",
|
||
"ROOT_ID": {
|
||
"id": "ROOT_ID",
|
||
"type": "ROOT",
|
||
"children": ["GRID_ID"],
|
||
},
|
||
"GRID_ID": {
|
||
"id": "GRID_ID",
|
||
"type": "GRID",
|
||
"children": [],
|
||
"parents": ["ROOT_ID"],
|
||
"meta": {"background": "BACKGROUND_TRANSPARENT"},
|
||
},
|
||
}
|
||
|
||
# Раскладываем dashboard вручную по 12-колоночной сетке:
|
||
# KPI в одну строку, затем аналитические блоки парами.
|
||
for chart in created_charts:
|
||
if chart:
|
||
chart_component_id = f"CHART-{chart['id']}"
|
||
layout = CHART_LAYOUT_BY_TITLE.get(
|
||
chart["title"],
|
||
{"width": 6, "height": 50, "x": 0, "y": 212},
|
||
)
|
||
positions[chart_component_id] = {
|
||
"id": chart_component_id,
|
||
"type": "CHART",
|
||
"children": [],
|
||
"parents": ["ROOT_ID", "GRID_ID"],
|
||
"meta": {
|
||
"chartId": chart['id'],
|
||
"sliceName": chart['title'],
|
||
"height": layout["height"],
|
||
"width": layout["width"],
|
||
"x": layout["x"],
|
||
"y": layout["y"],
|
||
},
|
||
}
|
||
positions["GRID_ID"]["children"].append(chart_component_id)
|
||
|
||
# Создаём дашборд
|
||
if created_charts:
|
||
try:
|
||
# Проверяем, существует ли дашборд
|
||
existing = db.session.query(Dashboard).filter_by(
|
||
slug=DASHBOARD_CONFIG["slug"]
|
||
).first()
|
||
|
||
if existing:
|
||
existing.description = DASHBOARD_CONFIG["description"]
|
||
existing.published = DASHBOARD_CONFIG["published"]
|
||
existing.json_metadata = metadata_json
|
||
existing.position_json = json.dumps(positions)
|
||
existing.slices = []
|
||
for chart_info in created_charts:
|
||
chart = db.session.query(Slice).filter_by(id=chart_info["id"]).first()
|
||
if chart:
|
||
existing.slices.append(chart)
|
||
db.session.commit()
|
||
logger.info(f"Dashboard '{DASHBOARD_CONFIG['dashboard_title']}' already exists (ID: {existing.id})")
|
||
logger.info("=" * 60)
|
||
logger.info("Dashboard already exists and metadata/layout were updated.")
|
||
logger.info(f"Dashboard URL: /superset/dashboard/{existing.id}/")
|
||
logger.info("=" * 60)
|
||
return True
|
||
|
||
# Создаём дашборд
|
||
dashboard = Dashboard(
|
||
dashboard_title=DASHBOARD_CONFIG["dashboard_title"],
|
||
slug=DASHBOARD_CONFIG["slug"],
|
||
description=DASHBOARD_CONFIG["description"],
|
||
published=DASHBOARD_CONFIG["published"],
|
||
json_metadata=metadata_json,
|
||
position_json=json.dumps(positions)
|
||
)
|
||
|
||
db.session.add(dashboard)
|
||
db.session.flush()
|
||
|
||
# Добавляем чарты к дашборду
|
||
for chart_info in created_charts:
|
||
if chart_info:
|
||
chart = db.session.query(Slice).filter_by(id=chart_info["id"]).first()
|
||
if chart:
|
||
dashboard.slices.append(chart)
|
||
|
||
db.session.commit()
|
||
|
||
logger.info(f"Created dashboard: {DASHBOARD_CONFIG['dashboard_title']} (ID: {dashboard.id})")
|
||
logger.info("=" * 60)
|
||
logger.info("Dashboard created successfully!")
|
||
logger.info(f"Dashboard URL: /superset/dashboard/{dashboard.id}/")
|
||
logger.info("=" * 60)
|
||
return True
|
||
|
||
except Exception as e:
|
||
logger.error(f"Failed to create dashboard: {e}")
|
||
import traceback
|
||
traceback.print_exc()
|
||
db.session.rollback()
|
||
return False
|
||
else:
|
||
logger.error("No charts created, cannot create dashboard")
|
||
return False
|
||
|
||
return False
|
||
|
||
if __name__ == "__main__":
|
||
sys.exit(0 if main() else 1)
|