Question

I need to delete all lines from a file containing a match for read (symbol), where (symbol) is any CJK character. In matches were read (symbol) is immediately preceded by A-Z or a-z, however, then the line should not be deleted. E.g., here are some sample lines and results:

Do you like to read books? (not deleted)
Can you read 书? ( deleted)
.read 书. (deleted)
This is some thread 线. (not deleted)

How can I delete only those lines matching (not A-Z or a-z)read (CJK symbol)?

Answer 1

I’m not entirely sure how to match CJK characters, but if you match non-ASCII characters you may achieve the results you’re looking for:

grep -vP "[^A-Za-z]read [\x80-\xFF]" file.txt

Theoretically, you should be able to do:

grep -vP "[^A-Za-z]read [\x{2E80}-\x{9FBB}]+" file.txt

However in my testing, I get error:

grep: character value in \x{...} sequence is too large

http://en.wikipedia.org/wiki/List_of_Unicode_characters#CJK_unified_ideographs

EDIT:

LC_ALL="POSIX" sed -r '/[^A-Za-z]read [\o200-\o377]+/d' file.txt

Results:

Do you like to read books? (not deleted)
This is some thread 线. (not deleted)

Also see:

How to delete all CJK text appearing immediately after a particular symbol?