Time for a little introduction to regular expressions:

Regular expressions (or "Regexps") are a tool for matching patterns in texts. They consist of
Ordinary characters just stand for themselves - "a" means "search for an a".

Metacharacters change the way, ordinary characters are interpreted. I will describe a few, but you are encouraged to research them, the web is full of articles about them.

[...]
defines a "character class". Any one character inside will be matched, but only one of them. Example:

Code:

a[bc]d

will match "abd" and "acd", but not "abcd" or "abbd", etc.. It is possible to reverse the meaning by using the caret "^" as the first character. This means every character NOT listed in the class:

Code:

a[^bc]d

This will match "axd" and "avd" - anything but "abd" and "acd". It is also possible to list consecutive characters by a "-": "[a.z]" means any (lowercase) character a-z, "[0-9]" means any digit 0-9. Notice, that i used "[ <t>]" (left bracket, space, tab, right bracket) in my expression above! That defines a character class which matches any whitespace, blanks OR tabs.


*

This acts as a multiplier to the expression before, meaning "zero or more of whatever the expression is". For example:

Code:

ab*c

means zero or more a's followed by a b. The strings "abc", "abbbbc" and even "ac" (zero or more!) are matched, but not "axc". If you combine this with the character classes you may get:

Code:

x[abc]*y

x, followed by any number of a's, b's or c's, followd by a y. It matches "xabcaababbbabby" and "xay", "xby" and "xcy" and even "xy" (again: zero or more), but not "xdy".


\( ... \)

This is a device for grouping and acts similar to brackets in math: it does nothing itself, but it groups together what is inside, for instance to make it manipulatable by the asterisk. Example:

Code:

\(abc\)*d

means any number of reduplications of the string "abc", followed by a "d". This matches "abcd" and "abcabcd", also "d", but not "abd", because before the "d" there is no "abc". Groupings can be nested.


\{n\} , \{m,n\}

This is also a multiplier for the expression before, like the asterisk, but limited by one or two numbers. The one-number variant means exactly that many reiterations, the two-number variant means between m and n reiterations. Example:

Code:

ab\{2\}c

is the same as "abbc". This one:

Code:

ab\{2,4\}c

matches "abbc", "abbbc" and "abbbbc" and nothing else. Again, this can be combined with other expressions, like classes. I used:

Code:

[0-9]\{14\}

to match exactly 14 digits. I could have written

Code:

[0-9][0-9][0-9][0-9][0-9][0-9][0-9][0-9][0-9][0-9][0-9][0-9][0-9][0-9]

for the same effect.


\

The backslash is part of some control constructs, as you have seen above. apart from that it is the "escape character". You have seen that several characters - metacharacters - are not meaning themselves but something different. The backslash strips this special meaning from them and makes them ordinary characters again. Like here:

Code:

ab\*c

Because the asterisk is escaped it is meant literal. This regexp matches the string "ab*c". Or this:

Code:

[0-9]\.[0-9]\{14\}

Usually the full stop is a metacharacter and means any one character. But here i want to use it literally and therefore escape it. The expression means a digit, followed by a full stop (decimal point), followed by eactly 14 other digits.

With this you should be able to decipher the sed script yourself.

I hope this helps.

bakunin