SQL Server 2008 generate script wizard gives me a script that results in “unclosed quotation marks”

Question

I have an SQL server 2008 database instance on one machine. Now I want to copy this database to another machine. I use the script wizard inside SQL Management Studio to generate a SQl-script with the schema and data. The script-file is rather big (around 17 GB).

Then I run the sql-script on the target machine it results in a :

Msg 105, Level 15, State 1 error with the message: Unclosed quotation mark after the character string

I do understand the problem of what unclosed quotation marks mean. But I don't understand why the error happens. Isn't the script generator able to handle quotations inside text strings like...hello, what's up...correctly and create a script that will escape such characters?

Is their a limit on the length of text for the script wizard? Is this causing the problem.

I don't want to and I cannot open the script-file in a text editor (too large, text editor will crash) and manually fix the problems.

Do you have any ideas?

Answer 1

Not a direct answer to the question but to duck this issue you could use one of the following other methods of copying the database to the new location.

1. Copying Databases with Backup and Restore
2. Using Detach and Attach

Method 1 is usually preferable as it keeps the source DB online and detaching can cause information held in the master database about the source to be lost (e.g. full text enabled status)

Answer 2

Solution for the SQL Server Import Issue

Pre-condition

In order to move the data from one SQL Server to another (e.g. from Production environment to Test environment) makes sense to use "Generate scripts" feature which is available in database options in SQL Server Management Studio. The result of this operation is text file with SQL commands that can be executed on another SQL Server. Usually these files are too big to execute them in SQL Server Management Studio, so we need to use sqlcmd command line utility from SQL Server installation package. In the most cases utility works smoothly and additional user actions are not necessary.

Issue description

In some rare cases the sqlcmd utility can fail with the import and raise the following error: "Unclosed quotation mark after the character string ..." which indicates that one of SQL queries has not been executed. This happens because sqlcmd works using stream processing, i.e. it reads some piece of data, processes it, reads next piece and so on. In some cases an input file can contain huge SQL instruction which size is bigger than the amount of the data that could be processed by sqlcmd at a time, so sqlcmd tries to execute broken SQL and fails.

Possible solutions

In order to fix this issue 2 approaches can be used:

• The sqlcmd utility can accept the "-a" parameter which defines the maximum size of packet (piece of data) that will be used during processing. The maximum value is 32767, the default value is 4096, so it makes sense to always use this parameter with maximum value.

  sqlcmd -i input.sql -a 32767 -o import_log.txt
  

• If the first approach didn't help and issue still appears, there is another, more difficult solution:

  • Install the Cygwin
    • During the installation, after some standard screens, stop on the screen "Select packages"
    • In "Search" field, enter "sed", and in the tree below expand the "Base" category and choose version not less than 4.2.2 for installation
    • Complete installation
    • Note: "sed" is the Linux utility which allows stream-based file processing
  • After installation is completed, run "Cygwin64 Terminal" from the desktop. We will use it for next steps

  • Go to the directory where the SQL file generated by SQL Server Management Studio is located. You need to use Linux style slashes "/" instead of Windows style which is "\"

      cd d:/temp
    

  • Change the encoding of the SQL file from UTF-16LE to UTF-8, because "sed" cannot process UTF-16LE, this conversion is safe for the data. The result will be a new file, that we will use in next step

      iconv -f UTF-16LE -t UTF-8 input.sql > input_utf8.sql
    

  • Convert the new file, to have one SQL query in one batch. The result will be a new file, that we will use in next step

      sed -e 's/^INSERT/GO\nINSERT/' input_utf8.sql > input_utf8_adapted.sql
    

  • Now the file "input_utf8_adapted.sql" should be processed by sqlcmd without any issues, so we can execute the following:

      sqlcmd -i input_utf8_adapted.sql -a 32767 -o import_log.txt
    

  • After execution is done, please check import_log.txt to make sure that no errors appeared

Answer 3

EDIT: Just noted from your comment that you're running sqlcmd -S server\database -i script.sql. There is a -I switch that stands for "Enable Quoted Identifiers". Try to run the command with this switch.

Btw, to edit a large file, consider using a nice editor like Notepad++ or UltraEdit. I wouldn't use a workstation without em :)

Answer 4

I ended up on this question after trying to find a solution to a similar problem I had. I also needed to dump a DB (with data) via Generate scripts wizard and the resulting file was too big to be executed from SSMS. So I tried the sqlcmd but ended with the error

Sqlcmd: Error: Syntax error at line 10 near command '"' in file 'script.sql'.

It turned out the cause of the issue was a record containing data with jQuery syntax in it - $(".someclass"). It's because it is also a way how to insert a variable into sqlcmd.

The solution is to disable variable substitution by adding -x command line argument.