sql: regular expression for certain pattern separation - sql

Question

In table "example" I have a column "col1" with following strings
some example text here x2.0.3-a abc
some other example text 1.5 abc
another example text 0.1.4 mnp
some other example text abc
another example text mnp
Now I need following things
Add the part before . to another column "col1"
Add the part . to another column "col2"
So the output should look like this
col1 col2
some example text here x2.0.3-a
some other example text 1.5
another example text 0.1.4
some other example text
another example text
Some of the properties of the string in col1 are
String in the col1 always end with either abc or mnp
Numbers like these x2.0.3-a or 0.1.4 are properties. These properties may not always exists in the col1 string. But if it exits then it always exists before the ending string abc or mnp.
there is always an space before properties and after the properties i.e another space between ending string abc/mnp and the properties.
So my question is how can I separate the properties and add them into col2?
One idea that comes into my head is that try to find something with *.* abc/mnp or *.*.* abc/mnp that is anything.anything. space abc/mnp OR anything.anything.anything space abc/mnp. I am not sure if I explained it properly.

Answer 1

As far as I understood, you'd like your column to be splitted into 3 columns. You should explain your 2nd column's scope and semantics better, so you can make sure to regular expression regularly matches with it.
I built a regex in parallel to the data you supplied, so it may not match for future incoming lines. Regex is here : https://regex101.com/r/seLgca/2/ What it does is, it captures three main groups:
(.+?)\s?([a-z]?\d(?:\.\d){1,2}(?:-[a-z])?)?\s(abc|mnp)
Let's break regex into parts:
(.+?)
\s?
([a-z]?\d(?:.\d){1,2}(?:-[a-z])?)?
\s
(abc|mnp)
Starting in reverse order, fifth part is simply matches abc or mnp. Forth part expects a space. Third part matches your second column if it exists, note that this part is for what you supplied so you can modify this part to suit your data better. Second part expects a space if it exists, this is for the lines contain empty second columns. First part is for the rest.
In Oracle, we have search and substring functions with regex as far as I know. So, you'll need a programming language to capture those groups.
I wrote a Java method for the purpose:
static List<String> getGroups(String content, String regex){
Pattern pattern = Pattern.compile(regex);
Matcher matcher = pattern.matcher(content);
List<String> groupsMatched = new ArrayList<String>();
if(matcher.find()){
for(int i=0; i<matcher.groupCount(); i++)
groupsMatched.add(matcher.group(i));
return groupsMatched;
}else
return null;
}
So, if I call the method with the lines you supplied like this:
for(String content : listOfContent){
List<String> groupsMatched = getGroups(content, regex);
if(groupsMatched != null)
System.out.println(groupsMatched.get(1) + "\t" + groupsMatched.get(2) + "\t" + groupsMatched.get(3) );
}
Here is what I have:
some example text here x2.0.3-a abc
some other example text 1.5 abc
another example text 0.1.4 mnp
some other example text null abc
another example text null mnp
Hope this helps.
Cheers,