在 Google Bigquery 中创建表 SQL 语法

Question

I've been reading the bigquery documentation since late last night and understand very little of it. It talks about loading data via different methods, but doesn't say how to create the table that I'm going to load data into. When I use the web UI it expects me to type out the schema. My table has over 400 columns. I will not type out hundreds of column names, types and lengths.

I've been uploading hundreds of GB of data in csv format to a google bucket. The csv files do not have column names. I have the schema in sql format which I prefer to use.

If I try creating a table through a query I get an error already on line 2 that says,

"Error: Encountered "" at line 2, column 1."

CREATE TABLE [example-mdi:myData_1.ST] (
`ADDRESS_ID` varchar(9),
`INDIVIDUAL_ID` varchar(2),
`FIRST_NAME` varchar(25),
`LAST_NAME` varchar(2),...

How can I do this or what is the right way?

Answer 1

You can use a CREATE TABLE statement to create the table using standard SQL. In your case the statement would look something like this:

CREATE TABLE `example-mdi.myData_1.ST` (
  `ADDRESS_ID` STRING,
  `INDIVIDUAL_ID` STRING,
  `FIRST_NAME` STRING,
  `LAST_NAME` STRING,
  ...
);

Answer 2

Mikhail is right and gets credit for the answer. If you're as slow as me you're going to want more details, because after he pointed me the right way it still took a while to figure out what he's talking about and how to get it done.

When you're at the create table user interface click the "edit as text" link.

In the text box input that pops up you'll enter something like:

ADDRESS_ID:string,
INDIVIDUAL_ID:string,
First_name:string,
Last_name:string...

Hyphens are not permitted.

Answer 3

When you create table in Web UI - you can enter schema field by field ( Edit as Fields mode - default mode) or you can enter schema as a text ( Edit as Text mode)
So, if you already have your schema in sql format you can just use it (you will might need to slightly adjust it to conform with BigQuery)

See more about creating tables with different clients (in We bUI section above option is not presented, so that's why you missed it I think)

PS As of today, BigQuery doe not support DDL - so CREATE TABLE is not available

Update

As of today - Jan 17, 2018 - BigQuerydata definition language support is now in Beta

Answer 4

As of April 2021, complex tables can be fully created directly using standard SQL (see the specific syntax guide ).

CREATE TABLE IF NOT EXISTS `project.dataset.table_name`
(
    someName STRING
    , dateTime TIMESTAMP NOT NULL -- REQUIRED or non-null column
    , index INT64 -- INT64 for INTEGER column
    , longitude FLOAT64 -- FLOAT64 for FLOAT column
    , arr ARRAY< -- declaring Array. This ARRAY is of datatype STRUCT
        STRUCT< -- declaring STRUCT
            a FLOAT64 -- the individual STRUCT members do not need the STRUCT column name again
            , b STRING
        >
    >
);

Answer 5

This is all new to me and I'm trying to insert a table and this is the code I'm using but I am getting the error code: Syntax error: Expected end of input but got keyword CREATE at [2:1]

Drop Table if exists new-project-369802.Covid.PercentPopulationVaccinated Create Table new-project-369802.Covid.PercentPopulationVaccinated ( Continent STRING(255), Location STRING(255), Date datetime, Population numeric, New_Vaccinations numeric, RollingPeopleVaccinated numeric ) INSERT INTO new-project-369802.Covid.PercentPopulationVaccinated SELECT dea.continent, dea.location, dea.date, dea.population, vac.new_vaccinations, SUM(cast(vac.new_vaccinations as int)) OVER(Partition BY dea.location ORDER BY dea.location, dea.date) AS RollingPeopleVaccinated FROM new-project-369802.Covid.covid_deaths AS dea JOIN new-project-369802.Covid.covid_vaccinations AS vac ON dea.location = vac.location and dea.date = vac.date WHERE dea.continent is not null

SELECT *,(rolling_count_vaccinated/population)*100 AS 
rolling_count_percentage
FROM `new-project-369802.Covid.PercentPopulationVaccinated`