Discussion on: Demystifying GREP

[Sérgio Araújo]

We can also use some advanced regex like in this example

any word preceded by big

grep -P '(?<=big)\s\w+' <<-EOF
big house
small house
big car
small car
EOF

outputs each line that NOT starts with big

grep -P '(?<!big)\s\w+' <<-EOF
big house
small house
big car
small car
EOF

The grep below will catch only lines that start with any word but not followed
by house

grep -P '\w+ (?!house)' <<-EOF
big house
small house
big car
small car
EOF

Now any word followed by house

grep -P '\w+ (?=house)' <<-EOF
big house
small house
big car
small car
EOF

The final example:

I had an html file with some wallpapers "name" on it, and I wanted to print just the names without full path

grep -Poh '(?<=href="/pt-br/wallpapers/details/)[^/]*' file-1.html

-P ...................... Perl regex
-o ...................... only the match
-h ...................... omit the filename
(?<=pattern) .... look behind
[^/]* ................... any character but slash

And an ASCII table about some of these regex patterns

    ┌─────────────┬──────────────────────┬────────────────────────────────┐
    │ Lookaround  │ Name                 │ What it Does                   │
    ├─────────────┼──────────────────────┼────────────────────────────────┤
    │ (?=foo)     │ Lookahead            │ word that follows is foo       │
    │ (?<=foo)    │ Lookbehind           │ preceding word is foo          │
    │ (?!foo)     │ Negative Lookahead   │ word that follows is NOT foo   │
    │ (?<!foo)    │ Negative Lookbehind  │ word behind is NOT foo         │
    └─────────────┴──────────────────────┴────────────────────────────────┘