将多个 Amazon Redshift 表合并到一个表中会出现错误：列 X 的类型为布尔值，但表达式的类型为字符变化提示：

Posted 2023-03-31

【中文标题】将多个 Amazon Redshift 表合并到一个表中会出现错误：列 X 的类型为布尔值，但表达式的类型为字符变化提示：

【英文标题】：Join multiple tables of Amazon Redshift in a single one obtains error: column X is of type boolean but expression is of type character varying Hint:

【发布时间】：2021-08-30 21:38:41

【问题描述】：

我正在尝试将多个 Amazon Redshift 表合并到一个表中。 初始表之一是这个：

create table order_customers(
  id int,
  email varchar(254),
  phone varchar(50),
  customer_id int,
  order_id int NOT NULL,
  ip text,
  geoip_location varchar(1024),
  logged_in boolean,
  PRIMARY KEY (id),
  FOREIGN KEY (order_id) REFERENCES orders (id)
);

我正在使用命令将数据插入到大表中：

INSERT INTO orders_large ( id, showid, created_at, status, status_enum, currency, tax_orders, shipping, discount_orders, 
  subtotal, total, store_id, payment_method_id, shipping_method_name, shipping_method_id, additional_information,
  payment_information, locale, shipping_required_orders, payment_method_type, coupons, payment_notification_id,     
  recover_token, updated_at, external, shipping_tax, shipping_discount, shipping_discount_decimal,  
  completed_at, payment_name, shipping_service_id, app_id, fulfillment_status, date_traffic_sources,
  landing_url,  referral_url, referral_code, utm_campaign, utm_source, utm_term, utm_medium, utm_content,       
  user_agent, subscription_id_traffic_sources, email, phone, customer_id_order_customers,    
  ip, geoip_location, logged_in, name, surname, company, address, street_number, city, postal, country, region,
  type, taxid, default_, region_format, municipality, latitude, longitude, subscription_id_addresses,
  customer_id_addresses, pickup_point_id, taxid_type, sku, qty, price, product_id, weight, 
  product_option_property_id, discount_order_products, shipping_required_orders_products, brand, 
  tax_order_products, width, height, length, volume, diameter, package_format)
SELECT o.id, o.showid, o.created_at, o.status, o.status_enum, o.currency, o.tax, o.shipping, o.discount, 
  o.subtotal, o.total, o.store_id, o.payment_method_id, o.shipping_method_name, o.shipping_method_id, o.additional_information,
  o.payment_information, o.locale, o.shipping_required, payment_method_type, coupons, payment_notification_id,     
  recover_token, o.updated_at, o.external, shipping_tax, o.shipping_discount, shipping_discount_decimal,  
  completed_at, payment_name, shipping_service_id, o.app_id, o.fulfillment_status, t.date,
  t.landing_url, t.referral_url, t.referral_code, t.utm_campaign, t.utm_source, t.utm_term, t.utm_medium, t.utm_content,       
  t.user_agent, t.subscription_id, oc.email, oc.phone, oc.customer_id, oc.order_id, oc.ip,
  oc.geoip_location, oc.logged_in, a.name, a.surname, a.company, a.address, a.street_number, a.city, a.postal,
  a.country, a.region, a.type, a.taxid, a.default_, a.region_format, a.municipality, a.latitude, a.longitude, a.order_id,
  a.subscription_id, a.customer_id, a.pickup_point_id, a.taxid_type, op.sku, op.qty, op.price, op.product_id,
  op.order_id, op.weight, op.product_option_property_id, op.discount, op.shipping_required, op.brand, 
  op.tax, op.width, op.height, op.length, op.volume, op.diameter, op.package_format
FROM  orders o
INNER JOIN traffic_sources t ON o.id = t.order_id
INNER JOIN order_customers oc ON  o.id = oc.order_id
INNER JOIN addresses a ON  o.id = a.order_id
INNER JOIN order_products op ON  o.id = op.order_id;

我收到此错误消息：

ERROR: column "logged_in" is of type boolean but expression is of type character varying Hint: You will need to rewrite or cast the expression.

我尝试在 oc.logged_in 字段中使用 DECODE(oc.logged_in, 'false', '0', 'true', '1')::varchar::bool ，但出现另一条错误消息：

ERROR: cannot cast type character varying to boolean

【问题讨论】：

你为什么要解码成一个字符串值然后试图将它强制转换成一个布尔值？我希望 DECODE(oc.logged_in, 'false', false, 'true', true, NULL) 或 DECODE(oc.logged_in, 'false', false, true) 会是您想要的。

两个选项都返回“错误：列“logged_in”是布尔类型，但表达式是字符变化类型提示：您需要重写或强制转换表达式。”

我也试过 DECODE(oc.logged_in, 'false', false, true)::Boolean, DECODE(oc.logged_in, 'false', false, 'true', true, NULL): :Boolean, DECODE(oc.logged_in, 'true', true, false)::Bool 并且错误总是相同的“错误：列“logged_in”是布尔类型，但表达式是字符类型不同提示：您将需要重写或转换表达式。”

【参考方案1】：

问题是字段之间的对应关系。它在使用“order_id”删除字段后工作。要在 Redshift 中投射，有 2 个选项：

CONVERT（类型、表达式） CAST（表达式 AS 类型）或表达式 :: 类型

字体：https://docs.aws.amazon.com/redshift/latest/dg/r_CAST_function.html#convert-function

【参考方案2】：

正如消息所说 - 列“logged_in”是布尔类型

因此，在您的 DECODE 中，您需要将其与布尔值进行比较，而不是字符串。试试：

DECODE(oc.logged_in, true, 'true', 'false')

The code above works for my understanding of you issue.  Below is test SQL which runs fine on Redshift.  

    create table oc as (select 1=1 as logged_in union all select 1=0);
    select * from oc;
    select DECODE(oc.logged_in, true, 'true string', 'false string') as test from oc;

I now expect that the issue is not in using oc.logged_in but rather with orders_large.logged_in and what you are putting in it.  What data type is logged_in defined as in orders_large?  Boolean, I assume.  Which should take a boolean value just fine w/o casting.  

Looking at your SQL I see that the number of elements in INSERT clause doesn't match the number of elements in the SELECT clause.  This mismatch is causing your SQL to try and put a different (text) value into orders_large.logged_in.  Here's a "diff" between the 2 lists (SELECT on the left / INSERT on the right):

    id                              id
    showid                              showid
    created_at                          created_at
    status                              status
    status_enum                         status_enum
    currency                            currency
    tax_orders                            | tax
    shipping                            shipping
    discount_orders                           | discount
    subtotal                            subtotal
    total                               total
    store_id                            store_id
    payment_method_id                       payment_method_id
    shipping_method_name                        shipping_method_name
    shipping_method_id                      shipping_method_id
    additional_information                      additional_information
    payment_information                     payment_information
    locale                              locale
    shipping_required_orders                      | shipping_required
    payment_method_type                     payment_method_type
    coupons                             coupons
    payment_notification_id                     payment_notification_id
    recover_token                           recover_token
    updated_at                          updated_at
    external                            external
    shipping_tax                            shipping_tax
    shipping_discount                       shipping_discount
    shipping_discount_decimal                   shipping_discount_decimal
    completed_at                            completed_at
    payment_name                            payment_name
    shipping_service_id                     shipping_service_id
    app_id                              app_id
    fulfillment_status                      fulfillment_status
    date_traffic_sources                          | date
    landing_url                         landing_url
    referral_url                            referral_url
    referral_code                           referral_code
    utm_campaign                            utm_campaign
    utm_source                          utm_source
    utm_term                            utm_term
    utm_medium                          utm_medium
    utm_content                         utm_content
    user_agent                          user_agent
    subscription_id_traffic_sources                   | subscription_id
    email                               email
    phone                               phone
    customer_id_order_customers                   | customer_id
                                      > order_id
    ip                              ip
    geoip_location                          geoip_location
    logged_in                           logged_in
    name                                name
    surname                             surname
    company                             company
    address                             address
    street_number                           street_number
    city                                city
    postal                              postal
    country                             country
    region                              region
    type                                type
    taxid                               taxid
    default_                            default_
    region_format                           region_format
    municipality                            municipality
    latitude                            latitude
    longitude                           longitude
    subscription_id_addresses                     | order_id
    customer_id_addresses                         | subscription_id
                                      > customer_id
    pickup_point_id                         pickup_point_id
    taxid_type                          taxid_type
    sku                             sku
    qty                             qty
    price                               price
    product_id                          product_id
                                      > order_id
    weight                              weight
    product_option_property_id                  product_option_property_id
    discount_order_products                       | discount
    shipping_required_orders_products                 | shipping_required
    brand                               brand
    tax_order_products                        | tax
    width                               width
    height                              height
    length                              length
    volume                              volume
    diameter                            diameter
    package_format                          package_format

As you can see there is an unmatched "orders_id" in the INSERT list just a couple of columns before logged_in.  You need to get the column alignment fixed.

【讨论】：

我尝试过并返回相同的错误"ERROR: column "logged_in" is of type boolean but expression is of type character varying Hint: You will need to rewrite or cast the expression." 我也尝试过DECODE(oc.logged_in, true, 'true', 'false')::Bool 并返回此错误"ERROR: cannot cast type text to boolean" 我在SELECT ... DECODE(oc.logged_in, true, 'true', 'false') ... 中使用它

您的第二次尝试将不起作用，因为您无法将字符串转换为布尔值或将布尔值转换为字符串。您可以将布尔值转换为 INT（变为 1 或 0）。我发布的代码有效，我将更新我的答案并提供更多详细信息。