This question already has answers here:
Can't we use [specifier] with like operator with Oracle?
(1 answer)
Query the list of CITY names starting with vowels (i.e., a, e, i, o, or u) from STATION
(18 answers)
Closed 3 years ago.
I am trying to find all the name that starts with either A,B or C in Oracle DB and I wrote this syntax:
SELECT NUME FROM JUCATORI
WHERE NUME LIKE '[ABC]%';
but it doesn't not give me any name (neither error), even though I am sure I have in my DB names that starts with either A,B or C.
LIKE doesn't work with character sets. (I think T-SQL extended LIKE to deal with such expressions, but Oracle's SQL doesn't.) You don't get an error, because LIKE '[ABC]%' looks for strings starting with an opening bracket, followed by an A, then a B, then a C, then a closing bracket
So either use OR:
SELECT nume FROM jucatori WHERE nume LIKE 'A%' OR nume LIKE 'B%' OR nume LIKE 'C%';
Or a regular expression:
SELECT nume FROM jucatori WHERE REGEXP_LIKE(nume, '^[ABC]');
One option is
where substr(nume, 1, 1) in ('A', 'B', 'C')
You're confusing regular expressions with 'like' conditions.
Like checks for exactly the string you pass to it, with the only three exceptions being:
1. % - any number of any characters
2. _ - any single character
3. ESCAPE clause that lets you define escape character
so you can use % as 'I'm looking for percent sign'
[ABC] meaning 'either A, B or C' is a regular expression syntax.
You may want to use regexp_like instead, or for such simple query just 'OR' several likes:
with examples as (
select 'Adam' ex from dual union all
select 'Dorothy' ex from dual union all
select '[ABC]something' ex from dual union all
select '[ABC]' from dual
)
select e.ex,
case when e.ex like '[ABC]%' then 'MATCHES' else null end as matches_like,
case when regexp_like(e.ex, '^[ABC].?') then 'MATCHES' else null end as matches_regexp_like,
case when e.ex like 'A%' or e.ex like 'B%' or e.ex like 'C%' then 'MATCHES' else null end as matches_complex_like
from examples e
Related
I'm looking at someone else's code which I do not have the option of running and can't figure out what the following REGEXP_LIKE is trying to match. Any help would be appreciated.
REGEXP_LIKE('field_name', '^(ABC:)?Z[DEF]')
What I think is happening is as follows but I think I am wrong:
Try to match any field that:
begins with ABC:
and ends D, E or F
I understand that the ^ matches the beginning of a string and that the () brackets group the expressions so is therefore grouping ABC:
However the ?Z is what is confusing me.
Any help would be appreciated, I can't seem to get my head around this no matter how many articles I read.
Try playing with some different strings:
with example as (select 'BC:ZDEF' as x from dual
union select 'ABC:D' from dual
union select 'ABC:ZE' from dual
union select 'ZE' from dual
union select 'ZF' from dual)
select x
from example
where REGEXP_like(x, '^(ABC:)?Z[DEF]');
Output:
x
ABC:ZE
ZE
ZF
So what's going on? You're right about ^ meaning the beginning of a line. The ? operator means the thing that comes before this is optional - it should occur 1 or 0 times. In this case, that's (ABC:), so that part of the string is optional.
Then we have a Z, which is mandatory, followed by a bracket expression, which means any single character listed between the brackets - so either D, E, or F.
So the expression means "a line starting with Z followed by D, E, or F, optionally with an "ABC:" at the beginning".
I'm new to Oracle and recently ran into the following query. I'm trying to understand what it's doing and hopefully rewrite it to optimize it. In this example, :NameList would be a comma separated list (like: "Bob,Bill,Fred") and then :N_NameList would be the number of tokens (in above example, 3)
SELECT ... FROM
(
SELECT
REGEXP_SUBSTR(:NameList,'[^,]+',1,LEVEL, 'i') Name
FROM DUAL CONNECT BY LEVEL <= :N_NameList
) x
INNER JOIN PEOPLE ppl
ON ppl.Name LIKE x.Name
...
From what I can tell, it expands out the delimited list into unique rows and then joins it with the following tables for each name, but I'm not sure if that's all it's doing. If that is the case, is there a better way to accomplish this?
You could try this instead:
select ...
from people ppl
where instr (','||:NameList||',',
','||ppl.name||',') > 0;
is there a better way to accomplish this?
Well, you could get rid of N_NameList because you can easily count number of tokens. This doesn't mean that it is a better way, it's just a different option. To be honest, it is probably slower option than yours as I have to calculate something that you entered as a parameter.
As this example is based on SQLPlus, I've used & instead of : for substitution variables. && means that it'll "remember" previously entered value (otherwise, I should type NameList twice.
Your current query:
SQL> select regexp_substr('&namelist', '[^,]+', 1, level, 'i') name
2 from dual
3 connect by level <= &n_namelist;
Enter value for namelist: Bob,Bill,Fred
Enter value for n_namelist: 3
Bob
Bill
Fred
Calculated N_NameList (using REGEXP_COUNT):
SQL> select regexp_substr('&&namelist', '[^,]+', 1, level, 'i') name
2 from dual
3 connect by level <= regexp_count('&&namelist', ',') + 1;
Enter value for namelist: Bob,Bill,Fred
Bob
Bill
Fred
A super simple example of my script looks as follows:
-- Report Name: "Report_1"
col letters new_value p_letters
SELECT letters
FROM param_table
WHERE report_name = 'Report_1';
CREATE TABLE temp_table_1
(letter varchar2(1));
INSERT INTO temp_table_1(letter)
SELECT DISTINCT letter
FROM table_alphabet
WHERE '&&p_letters' = '' OR letter IN (&&p_letters);
For some reason, our system has a table called param_table: users enter parameters through the UI, the parameters entered are written to param_table, and then my script pulls the user's parameters from param_table.
As far as I understand, the first SELECT statement selects the letters column from param_table and makes its values accessible in '&&p_letters'. In my INSERT INTO statement, when my WHERE clause looks like this...
WHERE letter IN (&&p_letters);
...and the user enters letters separated by single quotes, eg ('A', B', C'), the script runs fine. I want to make the parameter optional, so I adjusted the WHERE clause like this:
WHERE '&&p_letters' = '' OR letter IN (&&p_letters);
In my output file, I get the following error:
WHERE (('' = '') OR letter IN ()) *
ERROR at line ...:
ORA-00936: missing expression
The compiler has evaluated the substitution variable correctly as '', but I'm getting an error.
Any idea what I could be doing wrong here?
The ORA-00936 is because IN () is not valid - you're missing something inside that. It is that it is complaining about, not the '' = '' part, though the result of that is undefined. You can check both conditions:
SQL> select * from dual where '' = '';
no rows selected
SQL> select * from dual where dummy in ();
select * from dual where dummy in ()
*
ERROR at line 1:
ORA-00936: missing expression
If you set verify on you can see how the substitution is handled. For your original query you'd see:
old:INSERT INTO temp_table_1(letter)
SELECT DISTINCT letter
FROM table_alphabet
WHERE letter IN (&&p_letters)
new:INSERT INTO temp_table_1(letter)
SELECT DISTINCT letter
FROM table_alphabet
WHERE letter IN ('A','B','C')
3 rows inserted.
You can see that the post-substitution statement looks, and is, valid.
With your modified query you'd see:
old:INSERT INTO temp_table_1(letter)
SELECT DISTINCT letter
FROM table_alphabet
WHERE '&&p_letters' = '' OR letter IN (&&p_letters)
new:INSERT INTO temp_table_1(letter)
SELECT DISTINCT letter
FROM table_alphabet
WHERE ''A','B','C'' = '' OR letter IN ('A','B','C')
which generates an ORA-00920 because of the messed-up single quotes in the first expression. With no value from letters you'd instead see:
old:INSERT INTO temp_table_1(letter)
SELECT DISTINCT letter
FROM table_alphabet
WHERE '&&p_letters' = '' OR letter IN (&&p_letters)
new:INSERT INTO temp_table_1(letter)
SELECT DISTINCT letter
FROM table_alphabet
WHERE '' = '' OR letter IN ()
which is the error you saw, ORA-00936.
I'd be tempted to do this with a collection type, either your own, or if you're comfortable with it then a built-in one:
INSERT INTO temp_table_1(letter)
SELECT DISTINCT letter
FROM table_alphabet
WHERE SYS.DBMS_DEBUG_VC2COLL(&&p_letters) IS EMPTY
OR letter MEMBER OF SYS.DBMS_DEBUG_VC2COLL(&&p_letters);
That works with your three comma-separated values, or null, since an empty collection is allowed. Read more about is empty and member of.
It would be better, of course, to not store comma-separated lists in a single column value anyway, and to change your data model so this kind of manipulation and reliance on client behaviour isn't necessary.
Assuming you're stuck with the data model, you could at least avoid the client reliance buy tokenizing the string (I'm using one common approach below) and looking for matches. However, you also need to account for either the report name not being in the table at all or the report existing with no letters value, both of which are handled by the max(letters) .. is null check - which makes it a bit ugly.
It's all in one statement though, with no need for a separate query to get the parameters and no need for substitution variables. (And there may be better ways to do it!)
INSERT INTO temp_table_1 (letter)
SELECT DISTINCT letter
FROM table_alphabet
WHERE (
SELECT MAX(letters)
FROM param_table
WHERE report_name = 'Report_2'
) IS NULL
OR letter IN (
SELECT TRIM(q'[']' FROM REGEXP_SUBSTR(letters, '[^,]', 1, LEVEL))
FROM param_table
WHERE report_name = 'Report_2'
CONNECT BY REGEXP_SUBSTR(letters, '[^,]', 1, level) IS NOT NULL
);
Is there a fairly simple way to take an input parameter containing a comma seperated list of prefixes and return a cursor based on a select statement that uses these?
i.e. (Pseudocode)
PROCEDURE get_by_prefix(p_list_of_prefixes IN varchar2, r_csr OUT SYS_REFCURSOR)
IS
BEGIN
OPEN r_csr FOR
SELECT * FROM my_table where some_column LIKE (the_individual_fields_from p_list_of_prefixes ||'%')
END
I've tried various combinations, and now have two problems - coercing the input into a suitable table (I think it needs to go into a table type rather than a VARCHAR2_TABLE), and secondly getting the like clause to be effectively a SELECT from an internal 'pseudotable'...
EDIT: It seems that people are suggesting ways to use 'IN' with a set of potential values - whereas Im looking at using LIKE. I could use a similar technique - building up dynamic SQL, but was wondering if there isnt a more elegant way...
PL/SQL has no concept of a comma-separated list and no built-in splitter as in Perl etc, so you'll have to use one of the hand-rolled methods such as this one:
https://stewashton.wordpress.com/2016/08/01/splitting-strings-surprise
(Other methods are available.) Then it's just a matter of either populating a collection in one step and using it in the next, or else combining the two as something like this:
declare
p_list_of_prefixes varchar2(100) := 'John,Jim,Jules,Janice,Jenny';
begin
open :refcur for
with params as
( select x.firstname
from xmltable(
'ora:tokenize($X, "\,")'
passing p_list_of_prefixes as x
columns firstname varchar2(4000) path '.'
) x
)
, people as
( select 'Dave Clark' as fullname from dual union all
select 'Jim Potter' from dual union all
select 'Jenny Jones' from dual
)
select x.firstname, p.fullname
from params x
left join people p on p.fullname like x.firstname || '%';
end;
Output:
FIRSTNAME FULLNAME
-------------- -----------
John
Jim Jim Potter
Jules
Janice
Jenny Jenny Jones
Using LIKE the way you want is easy, but it is the wrong solution. (See my Comment under the original post).
Anyway - if by order of your superiors, or some other semi-legitimate reason, you must use a LIKE condition, it should look something like this:
... where ',' || p_list_of_whatever || ',' like '%,' || some_column || ',%
Concatenating commas at both ends of both sides of the comparison is needed, because you don't want Jo in the column to match John in the input list. Start from there and you will see why you need the commas on the right-hand side, and then follow from there and you will see why you need them on the left also.
I'm struggling with getting a query to work, and I could really use some help. We have an in house app that we use for building small web apps where I work. It's basically a drag and drop GUI. There's functionality built in to access query string values using the key.
I'm passing a comma separated list of values into a page through the query string. I'm then trying to use the list of values as part of an in clause in a query.
I can see that the value is correct in the query string.
orders=1,2,3
Here's the specific part of the query
"AND OrderNumber IN (this is where it maps from the query string)
I've tried running similar queries in Toad, and I think I've found the issue. It's giving an invalid number error, and I think it's wrapping the query string value in single quotes. I can replicate the error when I do "AND OrderNumber IN ('1,2,3')" in Toad.
Here's where I get really confused. The following works in Toad.
"AND OrderNumber IN ('1','2','3')"
So I tried recreating that by doing
select replace('1,2,3', ',', chr(39)||','||chr(39)) from dual;
I have confirmed that returns '1','2','3' in Toad.
However, I still get an Invalid Number error when I run the following in Toad.
AND OrderNumber IN (replace('1,2,3', ',', chr(39)||','||chr(39))
I've been racking my brain over this, and I can't figure it out. It seems to me that if "AND OrderNumber IN ('1','2','3')" works, and replace('1,2,3', ',', chr(39)||','||chr(39)) returns '1','2','3', that "AND OrderNumber IN (replace('1,2,3', ',', chr(39)||','||chr(39))" should work.
Any help you might be able to offer on this would be greatly appreciated. I know the rest of the query works. That's why I didn't post it. I'm just stuck on trying to get this IN clause working.
A change to phonetic_man's answer that will allow for NULL elements in the list. The regex format of '[^,]+' for parsing delimited lists does not handle NULL list elements and will return an incorrect value if one exists and thus its use should be avoided. Change the original by deleting the number 2 for instance and see the results. You will get a '3' in the 2nd element's position! Here's a way that handles the NULL and returns the correct value for the element:
SELECT TRIM(REGEXP_SUBSTR(str, '(.*?)(,|$)', 1, LEVEL, NULL, 1)) str
FROM ( SELECT '1,,3,4' str FROM dual )
connect by level <= regexp_count(str, ',') + 1;
See here for more info and proof: https://stackoverflow.com/a/31464699/2543416
Can you try the following query.
SELECT * FROM orders
WHERE orderno IN
(
SELECT TRIM(REGEXP_SUBSTR(str, '[^,]+', 1, LEVEL)) str
FROM ( SELECT '1,2,3,4' str FROM dual )
CONNECT BY INSTR(str, ',', 1, LEVEL - 1) > 0
)
The inline query splitting the string in different rows. So, on executing it you will get the following result.
SELECT trim(regexp_substr(str, '[^,]+', 1, LEVEL)) str
FROM ( SELECT '1,2,3,4' str FROM dual )
CONNECT BY instr(str, ',', 1, LEVEL - 1) > 0
1
2
3
4
Now, passing this result to the main query IN clause should work.
I think the desired clause to be built is:
AND OrderNumber IN (1,2,3)
A numeric list. The example you tested:
AND OrderNumber IN ('1','2','3')
works because an implicit conversion from a VARCHAR2 to a NUMBER is occurring for each element in the list.
The following clause will not work because no implicit conversion of the string '1,2,3' can be made (note the clause has a single string element):
AND OrderNumber IN ('1,2,3')
When doing a replace, you are converting the single string: 1,2,3 with the single string: 1','2','3 and this single string cannot be implicitly converted to a number.