![](/img/trans.png)
[英]Loading entire content of a csv file into a single column in a Hive table
[英]Loading CSV file on Hive Table with String Array
我正在嘗試將CSV文件插入到Hive中,其中一個字段是字符串數組。
這是CSV文件:
48,Snacks that Power Up Weight Loss,Aidan B. Prince,[Health&Fitness,Travel]
99,Snacks that Power Up Weight Loss,Aidan B. Prince,[Photo,Travel]
我嘗試創建這樣的表:
CREATE TABLE IF NOT EXISTS Article
(
ARTICLE_ID INT,
ARTICLE_NSAME STRING,
ARTICLE_AUTHOR STRING,
ARTICLE_GENRE ARRAY<STRING>
);
LOAD DATA INPATH '/tmp/pinterest/article.csv' OVERWRITE INTO TABLE Article;
select * from Article;
這是我得到的輸出:
article.article_id article.article_name article.article_author article.article_genre
48 Snacks that Power Up Weight Loss Aidan B. Prince ["[Health&Fitness"]
99 Snacks that Power Up Weight Loss Aidan B. Prince ["[Photo"]
它在最后一個字段article_genre中只取一個值。
誰能指出這里有什么問題?
幾個東西:
您缺少集合項的分隔符定義。
另外,我假設您希望you select * from article
語句中you select * from article
返回如下:
48 Snacks that Power Up Weight Loss Aidan B. Prince ["Health&Fitness","Travel"]
99 Snacks that Power Up Weight Loss Aidan B. Prince ["Photo","Travel"]
我可以給你一個例子,休息你可以擺弄它。 這是我的表定義:
create table article (
id int,
name string,
author string,
genre array<string>
)
row format delimited
fields terminated by ','
collection items terminated by '|';
這是數據:
48,Snacks that Power Up Weight Loss,Aidan B. Prince,Health&Fitness|Travel
99,Snacks that Power Up Weight Loss,Aidan B. Prince,Photo|Travel
現在做一個負載:
LOAD DATA local INPATH '/path' OVERWRITE INTO TABLE article;
並選擇語句來檢查結果。
最重要的一點 :
定義集合項的分隔符,不要強加你在正常編程中執行的數組結構。
此外,嘗試使字段分隔符與集合項分隔符不同,以避免混淆和意外結果。
為了在Hive表中插入字符串數組,我們需要注意以下幾點。
1. While creating Hive table.Collection items should be terminated by "," ('colelction.delim'=',',)
2. Data should be like that in CSV file
48 Snacks that Power Up Weight Loss Aidan B. Prince Health&Fitness,Travel
You can modify file by running below SED commands in follwing order:
- sed -i 's/\[\"//g' filename
- sed -i 's/\"\]//g' filename
- sed -i 's/"//g' filename
聲明:本站的技術帖子網頁,遵循CC BY-SA 4.0協議,如果您需要轉載,請注明本站網址或者原文地址。任何問題請咨詢:yoyou2525@163.com.