Using sed to clean up SchemaSync output - sed

Question

I have created a small sample of what I'm working with at this point. I am trying to properly quote the DEFAULT values, specifically #2 DEFAULT ONE value should be quoted as DEFAULT 'ONE':
#1 CREATE TABLE `table` (`column` int(10) unsigned DEFAULT NULL) ENGINE=InnoDB DEFAULT CHARSET=utf8mb4 COLLATE=utf8mb4_unicode_ci;
#2 ALTER TABLE `table` MODIFY COLUMN `column2` enum('ONE','TWO') NOT NULL DEFAULT ONE AFTER `column1`;
#3 ALTER TABLE `table` MODIFY COLUMN `column` varchar(64) NOT NULL DEFAULT '' FIRST;
I am currently using the following line against the above lines in order to fix line #2:
sed "s/DEFAULT \([a-zA-Z0-9_.]*\)/DEFAULT '\1'/g"
This is the output I'm getting:
#1 CREATE TABLE `table` (`column` int(10) unsigned DEFAULT 'NULL') ENGINE=InnoDB DEFAULT 'CHARSET'=utf8mb4 COLLATE=utf8mb4_unicode_ci;
#2 ALTER TABLE `table` MODIFY COLUMN `column2` enum('ONE','TWO') NOT NULL DEFAULT 'ONE' AFTER `column1`;
#3 ALTER TABLE `table` MODIFY COLUMN `column` varchar(64) NOT NULL DEFAULT '''' FIRST;
As you can see it fixes line #2 but #1 & #3 now have issues.
#1 DEFAULT 'NULL' (should remain DEFAULT NULL)
#1 DEFAULT 'CHARSET' (should remain DEFAULT CHARSET)
#2 DEFAULT 'ONE' (GOOD!)
#3 DEFAULT '''' (should remain DEFAULT '')
Is there a way to adjust the sed to ignore specific patterns such as DEFAULT NULL or DEFAULT CHARSET or DEFAULT ''?
sed "s/DEFAULT (not followed by NULL|CHARSET|'')\([a-zA-Z0-9_.]*\)/DEFAULT '\1'/g"
Or perhaps there is a better approach?
Thanks!

Answer 1

If Perl is your option, please try:
perl -pe "s/(DEFAULT) (?!(NULL|CHARSET|''))([a-zA-Z0-9_]+)/\1 '\3'/g" file
Output:
#1 CREATE TABLE `table` (`column` int(10) unsigned DEFAULT NULL) ENGINE=InnoDB DEFAULT CHARSET=utf8mb4 COLLATE=utf8mb4_unicode_ci;
#2 ALTER TABLE `table` MODIFY COLUMN `column2` enum('ONE','TWO') NOT NULL DEFAULT 'ONE' AFTER `column1`;
#3 ALTER TABLE `table` MODIFY COLUMN `column` varchar(64) NOT NULL DEFAULT '' FIRST;

Answer 2

Try:
sed -E "s/DEFAULT (NULL|CHARSET)/DEFAULT_\1/g; s/DEFAULT ([[:alnum:]_.]+)/DEFAULT '\1'/g; s/DEFAULT_(NULL|CHARSET)/DEFAULT \1/g" file
This works in three steps:
s/DEFAULT (NULL|CHARSET)/DEFAULT_\1/g
This hides the DEFAULT values that you don't want to change.
s/DEFAULT ([[:alnum:]_.]+)/DEFAULT '\1'/g
This changes the values that you do want to change.
Note that I changed the * to +. This means that empty strings will not be matched. This solves the problem in line #3.
I also changed [a-zA-Z0-9] to [:alnum:] so that the regex would match all alphanumeric characters in a unicode-safe way. (If this isn't what you wanted, just change this part back.)
s/DEFAULT_(NULL|CHARSET)/DEFAULT \1/g
This changes back the ones that you don't want to change.
This approach assumes that neither DEFAULT_NULL nor DEFAULT_CHARSET appear in your actual input. Based on what you have shown so far, this looks like a safe assumption.
Example
With your input file:
$ cat file
#1 CREATE TABLE `table` (`column` int(10) unsigned DEFAULT NULL) ENGINE=InnoDB DEFAULT CHARSET=utf8mb4 COLLATE=utf8mb4_unicode_ci;
#2 ALTER TABLE `table` MODIFY COLUMN `column2` enum('ONE','TWO') NOT NULL DEFAULT ONE AFTER `column1`;
#3 ALTER TABLE `table` MODIFY COLUMN `column` varchar(64) NOT NULL DEFAULT '' FIRST;
Our command produces:
$ sed -E "s/DEFAULT (NULL|CHARSET)/\n\1/g; s/DEFAULT ([[:alnum:]_.]+)/DEFAULT '\1'/g; s/\n(NULL|CHARSET)/DEFAULT \1/g" file
#1 CREATE TABLE `table` (`column` int(10) unsigned DEFAULT NULL) ENGINE=InnoDB DEFAULT CHARSET=utf8mb4 COLLATE=utf8mb4_unicode_ci;
#2 ALTER TABLE `table` MODIFY COLUMN `column2` enum('ONE','TWO') NOT NULL DEFAULT 'ONE' AFTER `column1`;
#3 ALTER TABLE `table` MODIFY COLUMN `column` varchar(64) NOT NULL DEFAULT '' FIRST;

Answer 3

With GNU awk for multi-char RS:
awk -v RS='DEFAULT \\w+' -v ORS= '
RT { split(RT,rt); if (rt[2] !~ /^(NULL|CHARSET)$/) RT=rt[1]" \047"rt[2]"\047" }
{ print $0 RT }
' file
#1 CREATE TABLE `table` (`column` int(10) unsigned DEFAULT NULL) ENGINE=InnoDB DEFAULT CHARSET=utf8mb4 COLLATE=utf8mb4_unicode_ci;
#2 ALTER TABLE `table` MODIFY COLUMN `column2` enum('ONE','TWO') NOT NULL DEFAULT 'ONE' AFTER `column1`;
#3 ALTER TABLE `table` MODIFY COLUMN `column` varchar(64) NOT NULL DEFAULT '' FIRST;