Regular Expression for Mixed cased words

Question

I want a regular expression that will help me identify if a certain pattern is present as a mixed case in a sentence or not.

Say:

Fox juMPed the rock

Now, I want to find that if the pattern jumped has any small letters in it or not. How can I achieve this?

Let me make it a little clearer. What I want is to make the keywords in a code file turn in UPPERCASE. For that I need to identify the keywords which are not already in uppercase.

Like Replaceshould ideally be REPLACE.

But to find such cases, I should be able to identify whether a word has any lowercase letter present or not.

Answer 1

printf %s\\n 'juMPEd'| sed 's/[^[:lower:]]//g'

OUTPUT

jud

Answer 2

Perl can do this fairly easily:

$ echo 'Fox juMPed the rock' | perl -pe 's/jumped/\U$&/i'
Fox JUMPED the rock

It looks for jumped, case insensitive, and replaces it with the uppercase version.

Answer 3

You can use grep -ioE to obtain all matches for a given regexp

$ echo "abc jumped def Jumped fgh JUMPED klm" > file.txt
$ grep -ioE "j\w+" file.txt > matches.txt
$ cat matches.txt
jumped
Jumped
JUMPED

Now, you can iterate on the result and remove full lower/upper matches (bash) :

$ for w in $(cat matches.txt); do if [[ ! $w == ${w^^} && ! $w == ${w,,} ]]; then echo $w; fi; done > mixed.txt
$ cat mixed.txt
Jumped

For each mixed word, you can invoke sed to uppercase the given match

$ for w in $(cat mixed.txt); do sed -i "s:$w:${w^^}:g" file.txt; done

Answer 4

Let me make it a little clearer. What I want is to make the keywords in a code file turn in UPPERCASE.

In that case, there is no need to test if a word has any lower case letters. It is easier and, in the end, likely more efficient to just replace them all.

$ echo Fox Jumped and JUMPED and juMPed the rock | sed 's/jumped/JUMPED/gi'
Fox JUMPED and JUMPED and JUMPED the rock

Refinement: Match only whole words

If you like, you can insist that only whole words be replaced:

sed 's/\<jumped\>/JUMPED/gi'

How to identify any occurrence of an identifier that is not fully upper case

Using grep -o

$ echo Fox Jumped and JUMPED and juMPed the rock | grep -woi jumped | grep '[[:lower:]]'
Jumped
juMPed

Two grep commands are used. The first does a case-insensitive search for the whole word jumped and the second restricts the matches to those with at least one lower case letter.

Using awk

This awk command will look for any occurrence of word in mixed-case or lower case:

awk -v 'RS=[^[:alnum:]]' 'tolower($0)=="jumped" && $0 ~ /[[:lower:]]/ {print $0}'

Here, words are defined to consist of alphanumeric characters. If your words allow more characters, adjust RS accordingly.

As an example:

$ echo Fox Jumped and JUMPED and juMPed the rock | awk -v 'RS=[^[:alnum:]]' 'tolower($0)=="jumped" && $0 ~ /[[:lower:]]/ {print $0}'
Jumped
juMPed

Answer 5

This can be done entirely within the shell, with a case construct.

string='Fox juMPed the rock'
case "$string" in
  *[:lower:]*) echo "The string contains lowercase letters.";;
esac

In ksh, you can use typeset -u to convert a string to uppercase. It doesn't matter whether the original was all-uppercase or not.

string='Fox juMPed the rock'
typeset -u string
echo "SHOUTING: $string"

If you want to process a file, you can use awk and its toupper function to convert a string to uppercase. You can test if s is already uppercase with s == toupper(s), though you might as well use toupper(s) regardless.