Shell Programming and Scripting

Hello,
Actually this is a follow-up of my earlier request to identify Sentence Boundaries while generating snippets for a search engine. The basic regex I have written to delimit sentence boundaries handles numbers and acronyms but I cannot get it to handle cases of
The full stops after Mr. Mrs. are automatically treated as sentence delimiters which is not desirable.
I tried the following syntax:

Code:

!(Dr\.|Mr\.|Mrs\.|Ms\.|[A-Z]\.|i\.e\.|w\.r\.t\.|e\.g\.|etc\.|viz\.)

to make the regex ignore a full-stop after such cases enumerated, but it does not work.
In fact the simple regex I had written has got murky and just does not perform any more.
Any help in correcting the regex would be appreciated.

Some sample sentences are given below:

Instead of writing things for a regex to not match, try getting something else in your regex to match it first. Regexes do greedy matching so whatever matches it first 'wins'.

What language is this regex for? This works in grep:

Code:

$ echo "Mr. Andrew visited me.  fleeb narf stuff." | egrep -o "([a-zA-Z]|(Mr|Ms|Dr|Mrs)[.]| )*[.]"
Mr. Andrew visited me.
  fleeb narf stuff.

$

A simplified example but hopefully conveys the idea.

Just a preference of mine, but I find it clearer to put special chars in [] than escape them to make them literal sometimes.

Many thanks. Works beautifully in egrep, but dies in Java. I wonder why. Does anybody know if Java demands a special regex set ?

regex really isn't the same everywhere. Might have been a good idea to post you were using java from the start.