Dockerfile

@@ -4,7 +4,7 @@ WORKDIR /app
 ### Pip
 USER root
 COPY ./requirements.txt /app/
-RUN pip config set global.index-url https://nexus.mujiannan.com:5001/repository/pypiserver/simple && \
+RUN pip3 config set global.index-url https://nexus.mujiannan.com:5001/repository/pypiserver/simple && \
     pip install -r requirements.txt
 USER 1001
 ## Copy files

doc-server.sh

@@ -1,2 +1,2 @@
 dbt docs generate
-dbt docs serve --host=0.0.0.0 --port=8080
+dbt docs serve

models/dw/dw.dim_date.sql

@@ -10,8 +10,7 @@
     ) 
 }}
 SELECT
-    toYYYYMMDD(date) AS date_id,
-    formatDateTimeInJodaSyntax(date, 'yyyyMMdd') AS date_id_str,
+    toUInt32(year(date) * 10000 + month(date) * 100 + day(date)) AS date_id,
     date AS full_date,
     toYear(date) AS year,
     toUInt8(quarter(date)) AS quarter,

models/dw/dw.dim_time.sql

@@ -2,16 +2,14 @@
     config(
         materialized='table',
         engine='MergeTree',
-        order_by='time_id',
+        order_by='time_id'
     ) 
 }}
-Select toYYYYMMDDhhmmss(`full_time`) As time_id
+Select cast(formatDateTime(`full_time`, '%Y%m%d%H%i') As Int64) As time_id
 	, `full_time`
 	, `date_id`
 	, hour(`full_time`) As `hour`
 	, minute(`full_time`) As `minute`
-	, formatDateTimeInJodaSyntax(`full_time`, 'yyyy-MM-dd HH:mm:ss') As full_time_str
-	, Cast(`date_id` As String) As date_id_str
 From(
 	SELECT 
 		arrayJoin(

models/dw/dw.fact_stock_daily.sql

@@ -2,12 +2,9 @@
     config(
         materialized='incremental',
         engine='MergeTree',
-        order_by='(date_id, ts_code)',
-        unique_key=['date_id', 'ts_code'],
+        order_by='ts_code, date_id',
+        unique_key=['ts_code', 'date_id'],
         incremental_strategy='delete+insert',
-        query_settings={
-            "join_algorithm": "'full_sorting_merge'",
-        }
     )
 }}
 Select 
@@ -38,7 +35,7 @@ Select
     StgDailyBasic.free_share As free_share,
     StgDailyBasic.total_mv As total_mv,
     StgDailyBasic.circ_mv As circ_mv
-From {{ source('finance', 'stg.tushare_daily') }} As StgDaily
+From {{ source('finance', 'stg.tushare_daily') }} As StgDaily Final
 Inner Join {{ ref('dw.dim_date') }} As DimDate
     On StgDaily.trade_date = Cast(DimDate.date_id As String)
 Any Left Join {{ source('finance', 'stg.tushare_daily_basic') }} As StgDailyBasic

models/dw/dw.fact_stock_minute.sql

@@ -1,40 +1,34 @@
 {{ 
     config(
-        materialized='incremental',
+        materialized='materialized_view',
         engine='MergeTree',
-        order_by='(time_id, ts_code)',
-        unique_key=['time_id', 'ts_code'],
-        incremental_strategy='delete+insert',
-        partition_by=['toYYYYMM(toDateTime(time_id))'],
-        query_settings={
-            "join_algorithm": "'full_sorting_merge'", 
-        }
+        order_by='ts_code, time_id',
+        unique_key=['ts_code', 'time_id'],
+        partition_by=['cast(floor(time_id / 1000000) As Int32)'],
+        query_settings={"join_algorithm": "'partial_merge'"}
     )
 }}
 
 Select 
-    OdsMinutes.ts_code As ts_code,
-    OdsMinutes.time_id As time_id,
-    OdsMinutes.close As close,
-    OdsMinutes.open As open,
-    OdsMinutes.high As high,
-    OdsMinutes.low As low,
-    OdsMinutes.vol As vol,
-    OdsMinutes.amount As amount,
-    OdsAdjFactor.adj_factor As adj_factor,
+    StgMinutes.ts_code As ts_code,
+    DimTime.time_id As time_id,
+    StgMinutes.close As close,
+    StgMinutes.open As open,
+    StgMinutes.high As high,
+    StgMinutes.low As low,
+    StgMinutes.vol As vol,
+    StgMinutes.amount As amount,
+    StgAdjFactor.adj_factor As adj_factor,
     now() As dt
-From {{ ref('ods.tushare_minutes') }} As OdsMinutes
-Any Left Join {{ ref('dw.dim_time') }} As DimTime
-    On OdsMinutes.time_id = DimTime.time_id
-Any Left Join {{ ref('ods.tushare_adj_factor') }} As OdsAdjFactor
-    On OdsMinutes.ts_code = OdsAdjFactor.ts_code
-        And OdsAdjFactor.date_id = DimTime.date_id
+From {{ source('finance', 'stg.tushare_minutes') }} As StgMinutes 
+Inner Join {{ ref('dw.dim_time') }} As DimTime
+    On StgMinutes.trade_time = formatDateTimeInJodaSyntax(full_time, 'yyyy-MM-dd HH:mm:ss')
+Any Left Join {{ source('finance', 'stg.tushare_adj_factor') }} As StgAdjFactor
+    On StgMinutes.ts_code = StgAdjFactor.ts_code
+        And StgAdjFactor.trade_date = Cast(DimTime.date_id As String)
 where 1 = 1
-{% if is_incremental() %}
-    And OdsMinutes.time_id >= (Select toYYYYMMDDhhmmss(toStartOfDay(YYYYMMDDhhmmssToDateTime(max(`time_id`)))) From {{ this }})
-{% endif %}
 {% if 'dev' in target.name %}
-    And OdsMinutes.time_id >= toYYYYMMDDhhmmss(dateAdd(Month,-1,today()))
+And StgMinutes.trade_time >= formatDateTimeInJodaSyntax(today(), 'yyyy-01-01 00:00:00')
 {% elif target.name == 'test' %}
-    And OdsMinutes.time_id >= toYYYYMMDDhhmmss(dateAdd(Year,-1,today()))
+And StgMinutes.trade_time >= formatDateTimeInJodaSyntax(dateAdd(Year,-1,today()), 'yyyy-01-01 00:00:00')
 {% endif %}

models/dw/schema.yml

@@ -2,27 +2,27 @@
 version: 2
 
 models:
-- name: dw.dim_exchange
+  - name: dw.dim_exchange
     description: "交易所"
     columns:
       - name: exchange_id
         description: "The primary key for this table"
-    data_tests:
+        tests:
           - unique
           - not_null
       - name: ts_exchange_code
         description: "交易所代码"
-    data_tests:
+        tests:
           - not_null
           - unique
       - name: exchange_name
         description: "交易所名称"
-    data_tests:
+        tests:
           - not_null
           - unique
-- name: dw.dim_date
+  - name: dw.dim_date
     description: "日期"
-  data_tests:
+    tests:
       - dbt_utils.unique_combination_of_columns:
           combination_of_columns: 
             - year
@@ -31,46 +31,41 @@ models:
     columns:
       - name: date_id
         description: "日期键"
-    data_tests:
+        tests:
           - unique
           - not_null
-  - name: date_id_str
-    description: "日期字符串（yyyyMMdd）"
-    data_tests:
-    - not_null
-    - unique
       - name: full_date
         description: "日期"
-    data_tests:
+        tests:
           - not_null
           - unique
       - name: year
         description: "年"
-    data_tests:
+        tests:
           - not_null
       - name: quarter
         description: "季度"
-    data_tests:
+        tests:
           - not_null
       - name: month
         description: "月"
-    data_tests:
+        tests:
           - not_null
       - name: day
         description: "日"
-    data_tests:
+        tests:
           - not_null
       - name: week_num_of_year
         description: "年中第几周"
-    data_tests:
+        tests:
           - not_null
       - name: day_num_of_week
         description: "星期"
-    data_tests:
+        tests:
           - not_null
-- name: dw.dim_time
+  - name: dw.dim_time
     description: "时间"
-  data_tests:
+    tests:
       - dbt_utils.unique_combination_of_columns:
           combination_of_columns: 
             - date_id
@@ -79,38 +74,29 @@ models:
     columns:
       - name: time_id
         description: "时间键"
-    data_tests:
+        tests:
           - unique
           - not_null
       - name: full_time
         description: "时间"
-    data_tests:
+        tests:
           - not_null
           - unique
       - name: date_id
         description: "日期键"
-    data_tests:
+        tests:
           - not_null
       - name: hour
         description: "小时"
-    data_tests:
+        tests:
           - not_null
       - name: minute
         description: "分钟"
-    data_tests:
+        tests:
           - not_null
-  - name: full_time_str
-    description: "时间字符串（yyyy-MM-dd HH:mm:ss）"
-    data_tests:
-    - not_null
-    - unique
-  - name: date_id_str
-    description: "日期字符串（yyyyMMdd）"
-    data_tests:
-    - not_null
-- name: dw.dim_hs_calendar
+  - name: dw.dim_hs_calendar
     description: "沪深交易日历"
-  data_tests:
+    tests:
       - dbt_utils.unique_combination_of_columns:
           combination_of_columns:
             - exchange_id
@@ -118,19 +104,19 @@ models:
     columns:
       - name: exchange_id
         description: "The primary key for this table"
-    data_tests:
+        tests:
           - not_null
       - name: date_id
         description: "交易日期键"
-    data_tests:
+        tests:
           - not_null
       - name: is_open
         description: "是否交易"
-    data_tests:
+        tests:
           - not_null
-- name: dw.dim_stock
+  - name: dw.dim_stock
     description: "股票"
-  data_tests:
+    tests:
       - dbt_utils.unique_combination_of_columns:
           combination_of_columns:
             - exchange_id
@@ -138,34 +124,34 @@ models:
     columns:
       - name: ts_code
         description: "tushare 股票编码"
-    data_tests:
+        tests:
           - unique
           - not_null
       - name: exchange_id
         description: "交易所键"
-    data_tests:
+        tests:
           - not_null
       - name: code
         description: "股票代码"
-    data_tests:
+        tests:
           - not_null
       - name: name
         description: "股票名称"
-    data_tests:
+        tests:
           - not_null
           - unique
       - name: full_name
         description: "股票全称"
-    data_tests:
+        tests:
           - not_null
           - unique
       - name: name_en
         description: "股票英文名称"
-    data_tests:
+        tests:
           - not_null
       - name: cn_spell
         description: "拼音缩写"
-    data_tests:
+        tests:
           - not_null
       - name: area
         description: "地区"
@@ -181,11 +167,11 @@ models:
         description: "退市日期"
       - name: is_active
         description: "当前是否在市"
-    data_tests:
+        tests:
           - not_null
       - name: is_hongkong_connect
         description: "是否沪深通标的"
-    data_tests:
+        tests:
           - not_null
       - name: actual_controller
         description: "实际控制人"
@@ -193,9 +179,9 @@ models:
         description: "实际控制人企业类型"
       - name: dt
         description: "更新时间"
-- name: dw.fact_stock_daily
+  - name: dw.fact_stock_daily
     description: "股票日数据"
-  data_tests:
+    tests:
       - dbt_utils.unique_combination_of_columns:
           combination_of_columns:
             - ts_code
@@ -203,11 +189,11 @@ models:
     columns:
       - name: ts_code
         description: "股票键"
-    data_tests:
+        tests:
           - not_null
       - name: date_id
         description: "日期键"
-    data_tests:
+        tests:
           - not_null
       - name: open
         description: "开盘价"
@@ -259,9 +245,9 @@ models:
         description: "总市值（万元）"
       - name: circ_mv
         description: "流通市值（万元）"
-- name: dw.fact_stock_minute
+  - name: dw.fact_stock_minute
     description: "股票分钟数据"
-  data_tests:
+    tests:
       - dbt_utils.unique_combination_of_columns:
           combination_of_columns:
             - ts_code
@@ -270,12 +256,12 @@ models:
       - name: ts_code
         description: "股票键"
         data_type: "String"
-    data_tests:
+        tests:
           - not_null
       - name: time_id
         description: "时间键"
         data_type: "Int64"
-    data_tests:
+        tests:
           - not_null
       - name: open
         description: "开盘价"
@@ -298,6 +284,8 @@ models:
       - name: adj_factor
         description: "复权因子"
         data_type: "Decimal(16,6)"
+        tests:
+          - not_null
       - name: dt
         description: "时间"
         data_type: "DateTime"

models/finance_source.yml

@@ -7,7 +7,7 @@ sources:
     - name: stg.tushare_trade_calendar
       description: >
         The source table for the trade calendar data from Tushare
-    data_tests:
+      tests:
       - dbt_utils.unique_combination_of_columns:
           combination_of_columns:
             - exchange
@@ -15,15 +15,15 @@ sources:
       columns:
       - name: exchange
         description: "The exchange code"
-      data_tests:
+        tests:
           - not_null
       - name: cal_date
         description: "The date"
-      data_tests:
+        tests:
           - not_null
       - name: is_open
         description: "Whether the exchange is open"
-      data_tests:
+        tests:
           - not_null
       - name: pretrade_date
         description: "The previous trading date"
@@ -31,7 +31,7 @@ sources:
       columns:
         - name: ts_code
           description: TS代码
-      data_tests:
+          tests:
             - not_null
         - name: symbol
           description: 股票代码
@@ -51,7 +51,7 @@ sources:
           description: 市场类型 （主板/中小板/创业板）
         - name: exchange
           description: 交易所代码
-      data_tests:
+          tests:
             - not_null
         - name: curr_type
           description: 交易货币
@@ -76,7 +76,7 @@ sources:
             - ts_code
     - name: stg.tushare_daily
       description: "沪深股票交易日线数据"
-    data_tests:
+      tests:
         - dbt_utils.unique_combination_of_columns:
             combination_of_columns:
               - ts_code
@@ -84,49 +84,49 @@ sources:
       columns:
         - name: ts_code
           description: TS代码
-      data_tests:
+          tests:
             - not_null
         - name: trade_date
           description: 交易日期
-      data_tests:
+          tests:
             - not_null
         - name: open
           description: 开盘价
-      data_tests:
+          tests:
             - not_null
         - name: high
           description: 最高价
-      data_tests:
+          tests:
             - not_null
         - name: low
           description: 最低价
-      data_tests:
+          tests:
             - not_null
         - name: close
           description: 收盘价
-      data_tests:
+          tests:
             - not_null
         - name: pre_close
           description: 昨收价
-      data_tests:
+          tests:
             - not_null
         - name: change
           description: 涨跌额
-      data_tests:
+          tests:
             - not_null
         - name: pct_chg
           description: 涨跌幅
-      data_tests:
+          tests:
             - not_null
         - name: vol
           description: 成交量
-      data_tests:
+          tests:
             - not_null
         - name: amount
           description: 成交额
     - name: stg.tushare_daily_basic
       description: "沪深股票每日指标数据"
-    data_tests:
+      tests:
         - dbt_utils.unique_combination_of_columns:
             combination_of_columns:
               - ts_code
@@ -189,7 +189,7 @@ sources:
           data_type: Nullable(Float32)
     - name: stg.tushare_adj_factor
       description: "沪深股票复权因子"
-    data_tests:
+      tests:
         - dbt_utils.unique_combination_of_columns:
             combination_of_columns:
               - ts_code
@@ -205,11 +205,11 @@ sources:
           description: "复权因子"
           data_type: Nullable(Decimal(16, 6))
         - name: dt
-      description: "更新时间"
+          description: "更新日期"
           data_type: DateTime
     - name: stg.tushare_minutes
       description: "沪深分钟级交易数据"
-    data_tests:
+      tests:
         - dbt_utils.unique_combination_of_columns:
             combination_of_columns:
               - ts_code
@@ -220,12 +220,12 @@ sources:
         - name: ts_code
           description: "The stock code."
           data_type: String
-      data_tests:
+          tests:
             - not_null
         - name: trade_time
           description: "The trading time."
           data_type: String
-      data_tests:
+          tests:
             - not_null
         - name: close
           description: "The closing price."

models/ods/ods.tushare_adj_factor.sql

@@ -1,19 +0,0 @@
-{{ 
-    config(
-        materialized='incremental',
-        engine="MergeTree",
-        order_by="(date_id, ts_code)",
-        unique_key=['date_id', 'ts_code'],
-        incremental_strategy='delete+insert',
-    )
-}}
-
-Select
-    ts_code As ts_code,
-    toYYYYMMDD(toDate(trade_date)) As date_id,
-    adj_factor As adj_factor,
-    dt As dt
-From {{ source('finance', 'stg.tushare_adj_factor') }} As StgAdjFactor
-{% if is_incremental() %}
-Where `trade_date` >= (Select formatDateTimeInJodaSyntax(YYYYMMDDToDate(max(`date_id`)), 'yyyy-MM-dd') From {{ this }})
-{% endif %}

models/ods/ods.tushare_minutes.sql

@@ -1,35 +0,0 @@
-{{ 
-    config(
-        materialized='incremental',
-        engine="MergeTree",
-        order_by='(time_id, ts_code)',
-        unique_key=['time_id', 'ts_code'],
-        partition_by=['toYYYYMM(toDateTime(time_id))'],
-        incremental_strategy='delete+insert',
-        query_settings={
-            "join_algorithm": "'full_sorting_merge'", 
-            "max_bytes_before_external_sort": "'1000M'",
-            "max_bytes_before_external_group_by":"'1000M'",
-        }
-    )
-}}
-
-Select 
-    StgMinutes.ts_code As ts_code,
-    toYYYYMMDDhhmmss(toDateTime(`StgMinutes`.`trade_time`)) As time_id,
-    StgMinutes.close As close,
-    StgMinutes.open As open,
-    StgMinutes.high As high,
-    StgMinutes.low As low,
-    StgMinutes.vol As vol,
-    StgMinutes.amount As amount
-From {{ source('finance', 'stg.tushare_minutes') }} As StgMinutes 
-where 1 = 1
-{% if is_incremental() %}
-    And StgMinutes.trade_time >= (Select formatDateTimeInJodaSyntax(toStartOfDay(YYYYMMDDhhmmssToDateTime(max(time_id))), 'yyyy-MM-dd HH:mm:ss') From {{ this }})
-{% endif %}
-{% if 'dev' in target.name %}
-    And StgMinutes.trade_time >= formatDateTimeInJodaSyntax(dateAdd(Month,-1,today()), 'yyyy-01-01 00:00:00')
-{% elif target.name == 'test' %}
-    And StgMinutes.trade_time >= formatDateTimeInJodaSyntax(dateAdd(Year,-1,today()), 'yyyy-01-01 00:00:00')
-{% endif %}

models/ods/schema.yml

@@ -1,54 +0,0 @@
-
-version: 2
-
-models:
-- name: ods.tushare_minutes
-  description: "沪深股票交易分钟线数据"
-  data_tests:
-  - dbt_utils.unique_combination_of_columns:
-      combination_of_columns:
-      - ts_code
-      - time_id
-  columns:
-  - name: ts_code
-    description: TS代码
-    data_tests:
-    - not_null
-  - name: time_id
-    description: 交易日期
-    data_tests:
-    - not_null
-  - name: close
-    description: 收盘价
-  - name: open
-    description: 开盘价
-  - name: high
-    description: 最高价
-  - name: low
-    description: 最低价
-  - name: vol
-    description: 成交量
-  - name: amount
-    description: 成交额
-  - name: dt
-    description: 更新时间
-- name: ods.tushare_adj_factor
-  description: "沪深股票复权因子"
-  data_tests:
-  - dbt_utils.unique_combination_of_columns:
-      combination_of_columns:
-      - ts_code
-      - date_id
-  columns:
-  - name: ts_code
-    description: TS代码
-    data_tests:
-    - not_null
-  - name: date_id
-    description: 交易日期
-    data_tests:
-    - not_null
-  - name: adj_factor
-    description: 复权因子
-  - name: dt
-    description: 更新时间

profiles.yml

@@ -15,7 +15,7 @@ finance_dbt:
       retries: 1
       compression: gzip
       connect_timeout: 10
-      send_receive_timeout: 24000
+      send_receive_timeout: 300
       cluster_mode: False
       use_lw_deletes: True
       check_exchange: True

requirements.txt

@@ -1 +1,2 @@
-dbt-clickhouse==1.8.0
+dbt==1.0.0.37.0
+dbt-clickhouse==1.7.3