[jon31]

Hey guys,

I'm doing a simple ereg_replace on a block of text, and I'd like to replace sets of words, not just the characters. example:

If I have the string "He said his head hurts", but I wanted to replace "He" with "That guy", I only want to replace the word "He", and not replace the letters "he" in the word "head". With my current code, the string of text would be translated to: "That guy said his that guyad hurts". See the problem?

Does that make sense? I have posted my code below.

$translate_txt = ereg_replace("He", "That guy", $translate_txt);

Any help would be greatly appreciated.

Jon Marus

[Turias]

I haven't tested this, but try:

preg_replace("/\bHe\b/", "That guy", $string);

[clam2000]

or you can make sure there are spaces arround it:
ereg_Replace(" He "," That guy ",$string);
(and what about returns... well a more general case would be this
ereg_replace("([[:space:]])He([[:space"]])","\\1That Guy\\2",$string);

how it works:
[[:space:]] = any white space (tab, space, return)
parentheses allow you to keep something, by refering to it later as \\# where the number is which set of perentheses it is.

--will

[Simon Mundy]

Originally posted by clam2000:
or you can make sure there are spaces arround it:
ereg_Replace(" He "," That guy ",$string);
(and what about returns... well a more general case would be this
ereg_replace("([[:space:]])He([[:space"]])","\\1That Guy\\2",$string);

how it works:
[[:space:]] = any white space (tab, space, return)
parentheses allow you to keep something, by refering to it later as \\# where the number is which set of perentheses it is.

--will

I think Turias' code is a bit more robust, as a '\b' word boundary is locale-sensitive, and not all words are seperated by whitespace

e.g.

"He's a funny old goat"

might need to be changed to

"That guy's a funny old goat"

- in this instance, the first word does not have whitespace, but the '\b' assertion will match the start of a line AND the ' character.

A bit picky of me, I know, but while it's subtly different it could make a huge impact on some search strings.

Cheers