Related
I'm running PSQL exports to CSV files for a few tables. They look like the below:
COPY table_name TO 'file_name' CSV
The issue is that some of these tables have text fields, in which the values for these fields contain both the delimiter (commas) and newlines. What would be the best way to do the export while removing the newlines across all columns?
Example table:
field1,field2,field3,field4
field1,field2,"field3, with, the delimiter",field4
field1,field2,"field3, with, the
delimiter and newline",field4
field1,"field2 with a
newline",field3,field4
How I'd want my export to look:
field1,field2,field3,field4
field1,field2,"field3, with, the delimiter",field4
field1,field2,"field3, with, the delimiter and newline",field4
field1,"field2 with a newline",field3,field4
Some solutions I've been considering:
Write a custom regex replace function and update the tables before I do the export.
See if there is a way to do the replace during the export transaction (is this possible?).
Perform the export as is and use another library/language to post-process the exported CSV.
Thanks for the help!
You can automatically compose an appropriate COPY statement with this SQL statement:
SELECT format(
'COPY (SELECT %s FROM %I.%I) TO ''filename'' (FORMAT ''csv'');',
string_agg(
format(
CASE WHEN data_type IN ('text', 'character varying', 'character')
THEN 'translate(%I, E''\n,'', '''')'
ELSE '%I'
END,
column_name
),
', '
ORDER BY ordinal_position
),
table_schema,
table_name
)
FROM information_schema.columns
WHERE table_schema = 'schema_name'
AND table_name = 'table_name'
GROUP BY table_schema, table_name;
If you are using psql, you can replace the final semicolon with \gexec to have psql run the resulting SQL statement for you in one go.
I have a table called Manufacturers which contains a column Name with a data type of Varchar(100) that has 5761 entries such as XYZ(XYZ Corp).
I wish to insert a space between the Z and the (.
I have seen the STUFF and LEFT commands but can't seem to figure out if they apply to my scenario.
You can try the REPLACE (Transact-SQL)
function as shown below.
update <yourTableName>
set <yourColumName> = replace(<yourColumName>, '(', ' ( ')
where <put the conditions here>
Here is an implementation to you.
create table test (Name varchar(20))
insert into test values ('XYZ(XYZ Corp)')
--selecting before update
select * from test
--updating the record
update test
set Name = replace(Name, '(', ' (')
where name like '%(%' --Here you can add the conditions as you want to restrict the number of rows to be updated based on the available data or the patterns.
--selecting after update
select * from test
Live Demo
I would recommend:
update Manufacturers
set name = replace(name, '(', ' (')
where name like '%[^ ](%';
This will only update rows where there is not already a space before the open paren.
Note: If you have multiple open parens, it will update all of them. If that is an issue, ask a new question with appropriate sample data.
I'm trying to use a select statement to get all of the columns from a certain MySQL table except one. Is there a simple way to do this?
EDIT: There are 53 columns in this table (NOT MY DESIGN)
Actually there is a way, you need to have permissions of course for doing this ...
SET #sql = CONCAT('SELECT ', (SELECT REPLACE(GROUP_CONCAT(COLUMN_NAME), '<columns_to_omit>,', '') FROM INFORMATION_SCHEMA.COLUMNS WHERE TABLE_NAME = '<table>' AND TABLE_SCHEMA = '<database>'), ' FROM <table>');
PREPARE stmt1 FROM #sql;
EXECUTE stmt1;
Replacing <table>, <database> and <columns_to_omit>
(Do not try this on a big table, the result might be... surprising !)
TEMPORARY TABLE
DROP TABLE IF EXISTS temp_tb;
CREATE TEMPORARY TABLE ENGINE=MEMORY temp_tb SELECT * FROM orig_tb;
ALTER TABLE temp_tb DROP col_a, DROP col_f,DROP col_z; #// MySQL
SELECT * FROM temp_tb;
DROP syntax may vary for databases #Denis Rozhnev
Would a View work better in this case?
CREATE VIEW vwTable
as
SELECT
col1
, col2
, col3
, col..
, col53
FROM table
You can do:
SELECT column1, column2, column4 FROM table WHERE whatever
without getting column3, though perhaps you were looking for a more general solution?
If you are looking to exclude the value of a field, e.g. for security concerns / sensitive info, you can retrieve that column as null.
e.g.
SELECT *, NULL AS salary FROM users
To the best of my knowledge, there isn't. You can do something like:
SELECT col1, col2, col3, col4 FROM tbl
and manually choose the columns you want. However, if you want a lot of columns, then you might just want to do a:
SELECT * FROM tbl
and just ignore what you don't want.
In your particular case, I would suggest:
SELECT * FROM tbl
unless you only want a few columns. If you only want four columns, then:
SELECT col3, col6, col45, col 52 FROM tbl
would be fine, but if you want 50 columns, then any code that makes the query would become (too?) difficult to read.
While trying the solutions by #Mahomedalid and #Junaid I found a problem. So thought of sharing it. If the column name is having spaces or hyphens like check-in then the query will fail. The simple workaround is to use backtick around column names. The modified query is below
SET #SQL = CONCAT('SELECT ', (SELECT GROUP_CONCAT(CONCAT("`", COLUMN_NAME, "`")) FROM
INFORMATION_SCHEMA.COLUMNS WHERE TABLE_NAME = 'users' AND COLUMN_NAME NOT IN ('id')), ' FROM users');
PREPARE stmt1 FROM #SQL;
EXECUTE stmt1;
If the column that you didn't want to select had a massive amount of data in it, and you didn't want to include it due to speed issues and you select the other columns often, I would suggest that you create a new table with the one field that you don't usually select with a key to the original table and remove the field from the original table. Join the tables when that extra field is actually required.
You could use DESCRIBE my_table and use the results of that to generate the SELECT statement dynamically.
My main problem is the many columns I get when joining tables. While this is not the answer to your question (how to select all but certain columns from one table), I think it is worth mentioning that you can specify table. to get all columns from a particular table, instead of just specifying .
Here is an example of how this could be very useful:
select users.*, phone.meta_value as phone, zipcode.meta_value as zipcode
from users
left join user_meta as phone
on ( (users.user_id = phone.user_id) AND (phone.meta_key = 'phone') )
left join user_meta as zipcode
on ( (users.user_id = zipcode.user_id) AND (zipcode.meta_key = 'zipcode') )
The result is all the columns from the users table, and two additional columns which were joined from the meta table.
I liked the answer from #Mahomedalid besides this fact informed in comment from #Bill Karwin. The possible problem raised by #Jan Koritak is true I faced that but I have found a trick for that and just want to share it here for anyone facing the issue.
we can replace the REPLACE function with where clause in the sub-query of Prepared statement like this:
Using my table and column name
SET #SQL = CONCAT('SELECT ', (SELECT GROUP_CONCAT(COLUMN_NAME) FROM INFORMATION_SCHEMA.COLUMNS WHERE TABLE_NAME = 'users' AND COLUMN_NAME NOT IN ('id')), ' FROM users');
PREPARE stmt1 FROM #SQL;
EXECUTE stmt1;
So, this is going to exclude only the field id but not company_id
Yes, though it can be high I/O depending on the table here is a workaround I found for it.
SELECT *
INTO #temp
FROM table
ALTER TABLE #temp DROP COlUMN column_name
SELECT *
FROM #temp
It is good practice to specify the columns that you are querying even if you query all the columns.
So I would suggest you write the name of each column in the statement (excluding the one you don't want).
SELECT
col1
, col2
, col3
, col..
, col53
FROM table
I agree with the "simple" solution of listing all the columns, but this can be burdensome, and typos can cause lots of wasted time. I use a function "getTableColumns" to retrieve the names of my columns suitable for pasting into a query. Then all I need to do is to delete those I don't want.
CREATE FUNCTION `getTableColumns`(tablename varchar(100))
RETURNS varchar(5000) CHARSET latin1
BEGIN
DECLARE done INT DEFAULT 0;
DECLARE res VARCHAR(5000) DEFAULT "";
DECLARE col VARCHAR(200);
DECLARE cur1 CURSOR FOR
select COLUMN_NAME from information_schema.columns
where TABLE_NAME=#table AND TABLE_SCHEMA="yourdatabase" ORDER BY ORDINAL_POSITION;
DECLARE CONTINUE HANDLER FOR NOT FOUND SET done = 1;
OPEN cur1;
REPEAT
FETCH cur1 INTO col;
IF NOT done THEN
set res = CONCAT(res,IF(LENGTH(res)>0,",",""),col);
END IF;
UNTIL done END REPEAT;
CLOSE cur1;
RETURN res;
Your result returns a comma delimited string, for example...
col1,col2,col3,col4,...col53
I agree that it isn't sufficient to Select *, if that one you don't need, as mentioned elsewhere, is a BLOB, you don't want to have that overhead creep in.
I would create a view with the required data, then you can Select * in comfort --if the database software supports them. Else, put the huge data in another table.
At first I thought you could use regular expressions, but as I've been reading the MYSQL docs it seems you can't. If I were you I would use another language (such as PHP) to generate a list of columns you want to get, store it as a string and then use that to generate the SQL.
Based on #Mahomedalid answer, I have done some improvements to support "select all columns except some in mysql"
SET #database = 'database_name';
SET #tablename = 'table_name';
SET #cols2delete = 'col1,col2,col3';
SET #sql = CONCAT(
'SELECT ',
(
SELECT GROUP_CONCAT( IF(FIND_IN_SET(COLUMN_NAME, #cols2delete), NULL, COLUMN_NAME ) )
FROM INFORMATION_SCHEMA.COLUMNS WHERE TABLE_NAME = #tablename AND TABLE_SCHEMA = #database
),
' FROM ',
#tablename);
SELECT #sql;
If you do have a lots of cols, use this sql to change group_concat_max_len
SET ##group_concat_max_len = 2048;
I agree with #Mahomedalid's answer, but I didn't want to do something like a prepared statement and I didn't want to type all the fields, so what I had was a silly solution.
Go to the table in phpmyadmin->sql->select, it dumps the query: copy, replace and done! :)
While I agree with Thomas' answer (+1 ;)), I'd like to add the caveat that I'll assume the column that you don't want contains hardly any data. If it contains enormous amounts of text, xml or binary blobs, then take the time to select each column individually. Your performance will suffer otherwise. Cheers!
Just do
SELECT * FROM table WHERE whatever
Then drop the column in you favourite programming language: php
while (($data = mysql_fetch_array($result, MYSQL_ASSOC)) !== FALSE) {
unset($data["id"]);
foreach ($data as $k => $v) {
echo"$v,";
}
}
The answer posted by Mahomedalid has a small problem:
Inside replace function code was replacing "<columns_to_delete>," by "", this replacement has a problem if the field to replace is the last one in the concat string due to the last one doesn't have the char comma "," and is not removed from the string.
My proposal:
SET #sql = CONCAT('SELECT ', (SELECT REPLACE(GROUP_CONCAT(COLUMN_NAME),
'<columns_to_delete>', '\'FIELD_REMOVED\'')
FROM INFORMATION_SCHEMA.COLUMNS
WHERE TABLE_NAME = '<table>'
AND TABLE_SCHEMA = '<database>'), ' FROM <table>');
Replacing <table>, <database> and `
The column removed is replaced by the string "FIELD_REMOVED" in my case this works because I was trying to safe memory. (The field I was removing is a BLOB of around 1MB)
You can use SQL to generate SQL if you like and evaluate the SQL it produces. This is a general solution as it extracts the column names from the information schema. Here is an example from the Unix command line.
Substituting
MYSQL with your mysql command
TABLE with the table name
EXCLUDEDFIELD with excluded field name
echo $(echo 'select concat("select ", group_concat(column_name) , " from TABLE") from information_schema.columns where table_name="TABLE" and column_name != "EXCLUDEDFIELD" group by "t"' | MYSQL | tail -n 1) | MYSQL
You will really only need to extract the column names in this way only once to construct the column list excluded that column, and then just use the query you have constructed.
So something like:
column_list=$(echo 'select group_concat(column_name) from information_schema.columns where table_name="TABLE" and column_name != "EXCLUDEDFIELD" group by "t"' | MYSQL | tail -n 1)
Now you can reuse the $column_list string in queries you construct.
I wanted this too so I created a function instead.
public function getColsExcept($table,$remove){
$res =mysql_query("SHOW COLUMNS FROM $table");
while($arr = mysql_fetch_assoc($res)){
$cols[] = $arr['Field'];
}
if(is_array($remove)){
$newCols = array_diff($cols,$remove);
return "`".implode("`,`",$newCols)."`";
}else{
$length = count($cols);
for($i=0;$i<$length;$i++){
if($cols[$i] == $remove)
unset($cols[$i]);
}
return "`".implode("`,`",$cols)."`";
}
}
So how it works is that you enter the table, then a column you don't want or as in an array: array("id","name","whatevercolumn")
So in select you could use it like this:
mysql_query("SELECT ".$db->getColsExcept('table',array('id','bigtextcolumn'))." FROM table");
or
mysql_query("SELECT ".$db->getColsExcept('table','bigtextcolumn')." FROM table");
May be I have a solution to Jan Koritak's pointed out discrepancy
SELECT CONCAT('SELECT ',
( SELECT GROUP_CONCAT(t.col)
FROM
(
SELECT CASE
WHEN COLUMN_NAME = 'eid' THEN NULL
ELSE COLUMN_NAME
END AS col
FROM INFORMATION_SCHEMA.COLUMNS
WHERE TABLE_NAME = 'employee' AND TABLE_SCHEMA = 'test'
) t
WHERE t.col IS NOT NULL) ,
' FROM employee' );
Table :
SELECT table_name,column_name
FROM INFORMATION_SCHEMA.COLUMNS
WHERE TABLE_NAME = 'employee' AND TABLE_SCHEMA = 'test'
================================
table_name column_name
employee eid
employee name_eid
employee sal
================================
Query Result:
'SELECT name_eid,sal FROM employee'
I use this work around although it may be "Off topic" - using mysql workbench and the query builder -
Open the columns view
Shift select all the columns you want in your query (in your case all but one which is what i do)
Right click and select send to SQL Editor-> name short.
Now you have the list and you can then copy paste the query to where ever.
If it's always the same one column, then you can create a view that doesn't have it in it.
Otherwise, no I don't think so.
I would like to add another point of view in order to solve this problem, specially if you have a small number of columns to remove.
You could use a DB tool like MySQL Workbench in order to generate the select statement for you, so you just have to manually remove those columns for the generated statement and copy it to your SQL script.
In MySQL Workbench the way to generate it is:
Right click on the table -> send to Sql Editor -> Select All Statement.
The accepted answer has several shortcomings.
It fails where the table or column names requires backticks
It fails if the column you want to omit is last in the list
It requires listing the table name twice (once for the select and another for the query text) which is redundant and unnecessary
It can potentially return column names in the wrong order
All of these issues can be overcome by simply including backticks in the SEPARATOR for your GROUP_CONCAT and using a WHERE condition instead of REPLACE(). For my purposes (and I imagine many others') I wanted the column names returned in the same order that they appear in the table itself. To achieve this, here we use an explicit ORDER BY clause inside of the GROUP_CONCAT() function:
SELECT CONCAT(
'SELECT `',
GROUP_CONCAT(COLUMN_NAME ORDER BY `ORDINAL_POSITION` SEPARATOR '`,`'),
'` FROM `',
`TABLE_SCHEMA`,
'`.`',
TABLE_NAME,
'`;'
)
FROM INFORMATION_SCHEMA.COLUMNS
WHERE `TABLE_SCHEMA` = 'my_database'
AND `TABLE_NAME` = 'my_table'
AND `COLUMN_NAME` != 'column_to_omit';
I have a suggestion but not a solution.
If some of your columns have a larger data sets then you should try with following
SELECT *, LEFT(col1, 0) AS col1, LEFT(col2, 0) as col2 FROM table
If you use MySQL Workbench you can right-click your table and click Send to sql editor and then Select All Statement This will create an statement where all fields are listed, like this:
SELECT `purchase_history`.`id`,
`purchase_history`.`user_id`,
`purchase_history`.`deleted_at`
FROM `fs_normal_run_2`.`purchase_history`;
SELECT * FROM fs_normal_run_2.purchase_history;
Now you can just remove those that you dont want.
I'm trying to insert post codes into my database but getting rid of the underscores.
I have a table called FeedDataSetMapping that is used to map the fields before they get inserted:
INSERT INTO FeedDataSetMapping (
[source_field]
,[database_field]
,[template_id]
,[conversion_id]
,[order_id]
,[values_group]
,[direct_value]
,[value_regex]
,[condition_regex]
,[split_separator]
,[enclosing_character]
,[cumulative_field]
,[cumulative_format])
VALUES
('manufacturerId','manufacturer_Id',#template_id,0,0,null,null,null,null,null,null,null,null),
('dealership','leasing_broker_name',#template_id,0,0,null,null,null,null,null,null,null,null),
('manufacturersDealerId','supplier_ref',#template_id,0,0,null,null,19,null,null,null,null,null),
('address1','address1',#template_id,0,0,null,null,null,null,null,null,null,null),
('address2','address2',#template_id,0,0,null,null,null,null,null,null,null,null),
('postcode','post_code',#template_id,0,0,null,null,null,null,null,null,null,null),
('telephone','telephone',#template_id,0,0,null,null,null,null,null,null,null,null),
('fax','fax_number',#template_id,0,0,null,null,null,null,null,null,null,null),
('email','email',#template_id,0,0,null,null,null,null,null,null,null,null),
('website','web_address',#template_id,0,0,null,null,null,null,null,null,null,null),
('NewCarSales','service_mask',#template_id,0,0,null,1,null,'^(?!(?i:^0$|^n$|^no$|^f$|^false$|^$))',null,null,1,null),
('UsedCarSales','service_mask',#template_id,0,0,null,2,null,'^(?!(?i:^0$|^n$|^no$|^f$|^false$|^$))',null,null,1,null),
('Servicing','service_mask',#template_id,0,0,null,8,null,'^(?!(?i:^0$|^n$|^no$|^f$|^false$|^$))',null,null,1,null),
('Repairs','service_mask',#template_id,0,0,null,16,null,'^(?!(?i:^0$|^n$|^no$|^f$|^false$|^$))',null,null,1,null),
('Longitude','longitude',#template_id,0,0,null,null,null,null,null,null,null,null),
('Latitude','latitude',#template_id,0,0,null,null,null,null,null,null,null,null)
This already contains some condition regex that in case that this field contains some text it converts it to true or false respectively.
What I need is a condition_regex that gets rid of these underscores and replaces it with a blank space i.e: 'GDB_A45' to 'GDB A45'. I don't know much about regex so any idea would be greatly appreciated. Thanks in advance!
SQL Server does not have much of regular expression support, but in this case I don't think you need it. You can do a simple replace:
UPDATE mytable
SET mycolumn = REPLACE(mycolumn, '_', ' ')
WHERE mycolumn LIKE '%[_]%'
To do this while updating you can use INSERT ... SELECT instead of INSERT ... VALUES:
INSERT INTO mytable (mycolumn)
SELECT REPLACE('my data 1', '_', ' ') UNION
SELECT REPLACE('my data 2', '_', ' ') UNION
SELECT REPLACE('my_data_3', '_', ' ') UNION
...
There will be some maximum number of unions you can do, so you should split your inserts into batches with this method.
Or, you could define a trigger on the target table that will do the job for you:
CREATE TRIGGER mytrigger ON mytable
AFTER INSERT AS
BEGIN
UPDATE mytable
SET mytable.mycolumn = REPLACE(i.mycolumn, '_', ' ')
FROM mytable
INNER JOIN inserted i
ON i.id = mytable.id
AND i.mycolumn LIKE '%[_]%'
END
... where it is assumed your table has a primary key named id.
Well after been thinking a while I got to the conclusion that would be easier if I replace the underscore from the scraped data during the scraping (in the c# code) before I generate the XML file. That would avoid me a lot of headaches. Anyway thank you for your help guys ;)
I need to concatenate a variable table name into my SQL query such as the following...
ALTER FUNCTION fn_myfunction(#KeyValue text)
BEGIN
INSERT INTO #tmpTbl1
SELECT #KeyValue AS fld1
FROM tbl + #KeyValue + KeyValue.fld1
I also attempted the following but it told me I had to declare the table variable?
ALTER FUNCTION fn_myfunction(#KeyValue text, #KeyTable text)
FROM #KeyTable.fld1
You can accomplish this (if I understand what you are trying to do) using dynamic SQL.
The trick is that you need to create a string containing the SQL statement. That's because the tablename has to specified in the actual SQL text, when you execute the statement. The table references and column references can't be supplied as parameters, those have to appear in the SQL text.
So you can use something like this approach:
SET #stmt = 'INSERT INTO #tmpTbl1 SELECT ' + #KeyValue
+ ' AS fld1 FROM tbl' + #KeyValue
EXEC (#stmt)
First, we create a SQL statement as a string. Given a #KeyValue of 'Foo', that would create a string containing:
'INSERT INTO #tmpTbl1 SELECT Foo AS fld1 FROM tblFoo'
At this point, it's just a string. But we can execute the contents of the string, as a dynamic SQL statement, using EXECUTE (or EXEC for short).
The old-school sp_executesql procedure is an alternative to EXEC, another way to execute dymamic SQL, which also allows you to pass parameters, rather than specifying all values as literals in the text of the statement.
FOLLOWUP
EBarr points out (correctly and importantly) that this approach is susceptible to SQL Injection.
Consider what would happen if #KeyValue contained the string:
'1 AS foo; DROP TABLE students; -- '
The string we would produce as a SQL statement would be:
'INSERT INTO #tmpTbl1 SELECT 1 AS foo; DROP TABLE students; -- AS fld1 ...'
When we EXECUTE that string as a SQL statement:
INSERT INTO #tmpTbl1 SELECT 1 AS foo;
DROP TABLE students;
-- AS fld1 FROM tbl1 AS foo; DROP ...
And it's not just a DROP TABLE that could be injected. Any SQL could be injected, and it might be much more subtle and even more nefarious. (The first attacks can be attempts to retreive information about tables and columns, followed by attempts to retrieve data (email addresses, account numbers, etc.)
One way to address this vulnerability is to validate the contents of #KeyValue, say it should contain only alphabetic and numeric characters (e.g. check for any characters not in those ranges using LIKE '%[^A-Za-z0-9]%'. If an illegal character is found, then reject the value, and exit without executing any SQL.
You could make use of Prepared Stements like this.
set #query = concat( "select name from " );
set #query = concat( "table_name"," [where condition] " );
prepare stmt from #like_q;
execute stmt;