Convert a paragraph into sentences with dynamic memory

Question

How can I convert a paragraph into sentences? I have a function signature as follows:

char **makeSentences(char *paragraph);

In which:

paragraph is a string containing several sentences. Paragraph ensures that each sentence ends with a period (.) and the whole paragraph ends with a null-terminator.
returns a dynamically allocated array of sentences.

I'm a bit confused of how to allocate the memory for the array of sentences on the fly.

How is caller supposed to know how many sentences will be returned by the function? — JosephH, Nov 20 '11 at 08:07
Your question is tagged with `malloc` and `realloc`. Have you tried to use these facilities? — n. m. could be an AI, Nov 20 '11 at 08:14
It's all to do with malloc and realloc, but I still have no clue how can I allocate memory (for a new chacter or a sentence) on the fly :) — antiopengl, Nov 20 '11 at 08:18
I meant how is **caller**, not callee supposed to know how many sentences will be returned by the function? — JosephH, Nov 20 '11 at 08:21
Not all sentences end with a period. Do you realise that some sentences end with a question mark? Hell, some questions even end with an exclamation mark! — wildplasser, Nov 20 '11 at 11:40

score 1 · Answer 1 · answered Nov 20 '11 at 08:20

Within makeSentences you will need to examine paragraph to determine the number of characters each sentence contains, including the period and a character for \0.

You can allocate the memory like this:

int i;
char** sentences = (char**)malloc( sizeof(char*) * number_of_sentences );
for (i = 0; i < number_of_sentences; i++) {
    sentences[i] = (char*)malloc( sizeof(char) * length_of_sentences[i] );
}

where length_of_sentences is an array containing the sentence lengths you've discovered.

But how do I manage to copy each character into the sentence array? — antiopengl, Nov 20 '11 at 11:38

score 0 · Answer 2 · answered Nov 20 '11 at 08:19

Write a helper method to provide a count for the number of paragraphs in advance, and then malloc an array of char* pointers based on this number. It will be something like this although the below code is untested so it might not work straight off:

int getSentenceCount( char* paragraph) {
   int sentenceCount = 0, i = 0;
   for(i=0; i < sizeof(paragraph); i++){
        if(paragraph[i] == '.') sentenceCount++;
   }

   return sentenceCount;
}

Once you've got a sentence count the rest should be easy using string utility functions (See string.h)

score 0 · Answer 3 · answered Nov 20 '11 at 10:23

So that the caller can determine when it reached the end of the array, just make the array one entry larger and use NULL for the last/extra entry.

The hard part is definitely going to be determining where sentences start/end. Here's some example sentences (one sentence per line):

Hello world!
Are you hungry?
She said "Hello. I'm Sara!" and smiled.
I installed version 1.3.
They laughed at the F.B.I. and ran.
The function returned the sequence 1, 3, 5, ..., 33.
7 is a lucky number (and this sentence doesn't start with a capital letter).

Put all these sentences into a single paragraph to see if your "end of sentence" detection works correctly.

Convert a paragraph into sentences with dynamic memory

3 Answers3