extracting a sentence

[serhannn]

Hello everybody,

I have a paragraph containing many sentences and I need to extract a certain sentence, which contains a certain. I have no problem finding the word with my code but my question is, how can I extract the whole sentence? The code should work, regardless of the position of the word in the sentence. Here I give an example:

A new crisis is emerging, a global food catastrophe that will reach further and be more crippling than anything the world has ever seen. The credit crunch and the reverberations of soaring oil prices around the world will pale in comparison to what is about to transpire, Donald Coxe, global portfolio strategist at BMO Financial Group said at the Empire Club's 14th annual investment outlook in Toronto on Thursday.

For instance, I search for the word catastrophe in this text and I find it. Now I need to extract the sentence:"A new crisis is emerging, a global food catastrophe that will reach further and be more crippling than anything the world has ever seen. "
I thought that I could use string.find until the characters ". " to get to the end of the sentence, but I also need to retrieve the part which comes before the word catastrophe. I would appreciate your ideas and help over this topic.

Thank you very much.

[ddanbe]

Proceed in the same way. Find the previous '.', then move forward until the first capital.

[serhannn]

Proceed in the same way. Find the previous '.', then move forward until the first capital.

But I can't decide whether it's beginning of a sentence looking at the capital. It can also be a name or another things.

[ddanbe]

Come to think of it, it is not evident.
What would you do with a sentence like: This person holds a Ph.D., but this is not really a C++ problem any more.

[serhannn]

In a loop, I used this code to start from the found word and take until the first encounter with a dot. But I think there's something wrong with it, because in output there are some errors. I search in text files, which are actually source codes of some webpages, so they contain many HTML tags. They also interrupt. Is there any better algorithm to avoid that?

size_t pos1,pos2;
pos1=line.find(keyword_vector[i]); 
pos2=line.find(".",pos1); 	
string sentence = line.substr(pos1,pos2);

Thanks for help.

[Freaky_Chris]

Perhaps you should read the entire file in, strip out all html tags as you do so, that way you are left with just text. The go over and process what you have for sentences. But as ddanbe said, this isn't really a job for C++

The closest you will get to checking for the start and end of a sentence is [dot][space][capital] you will need to use that to check for both the beginning and end.

Chris

Chris

[Comatose]

You still have to take into account that reading in a file has newlines. So, what if my sentence is nearing the end of the line, and I need to break it into the next line? If "word wrap" isn't used, but instead manual line breaks, sentences could be span on multiple lines :/