RLIKE 表达式给出“不匹配的输入”错误

Question

I am querying a Hive table through Redash. I have a query similar to the following:

SELECT 
    CAST(id AS INT) as id,
    COUNT(sales) AS num_sales
FROM 
    sales_table
WHERE 
    id RLIKE '\d*'
GROUP BY 
    id

I am trying to select only IDs which are a string of digits, hence the WHERE clause. This is giving me this error:

Error running query: line 42:20: mismatched input 'RLIKE'. Expecting: '%', '*', '+', '-', '.', '/', 'AT', '[', '||', <expression>

Answer 1

You can try:

WHERE id RLIKE '^[0-9]+$'

Note that the above expression ensures that id is made of digits only (and non-empty). You original regexp was checking if id contained 0 to n digits (that's what the * quantifier means): this is way to permissive, as it would basically allow anything that is not null.

Answer 2

1)

Hive regex requires double escaping

also, for full match you should use anchoring, therefore -

id RLIKE '^\\d+$'

if empty string is also an option then -

id RLIKE '^\\d*$'

2)
You can take the opposite approach and search for strings that does not contain a non-digit -

id  NOT RLIKE '\\D'

3)
In Hive a false casting results in null (and not an exception) therefore you could use -

cast(id as int) is not null

or

int(id) is not null

Answer 3

In Hive, the code should compile with either REGEXP or RLIKE . However, your regular expression does not do what you want. You want:

WHERE id RLIKE '^[0-9]*$'

That is, all characters are digits, rather than it just containing a single digit. Well, actually, your version would match any string, because the digit is optional.