Search For Subtext In Text With A Defined Algorithm - c++

Question

I want to create a program to search for a subtext in a text.
For example, I have this text: abcdeabbdfeg
And in that text I want to find: cd
But I want to use this algorithm:
start = 1
end = string length of the text
middle = (start + end) / 2
if (pattern < text[middle]) end = mid - 1;
if (pattern > text[middle]) start = mid + 1;
...and continue until the pattern is found in the text
So, I already have a simple program that completely works without any problem but without that algorithm above, so now I only want to implement that algorithm above in my program, I have tried many ways, but my program won't show anything in any case, after I add that algorithm...
This is the code that I have and works:
void search(char *pat, char *txt)
{
int M = strlen(pat);
int N = strlen(txt);
for (int i = 0; i <= N - M; i++)
{
int j;
for (j = 0; j < M; j++)
{
if (txt[i+j] != pat[j])
break;
}
if (j == M)
{
printf("Pattern found at index %d \n", i);
}
}
}
And this is the code above with the implementation of the algorithm:
int _tmain(int argc, _TCHAR* argv[])
{
char t[32];
cout << "Please enter your text (t):";
cin >> t;
char p[32];
cout << "Please enter the pattern (p) you wish to look for in that text (t):";
cin >> p;
int start, end = 0;
double middle = 0;
start = 1;
end = strlen(t);
while (start <= end)
{
int M = strlen(p);
int N = strlen(t);
middle = std::ceil((start + end) / 2.0);
int mid = (int)middle;
for (int i = mid; i <= M; i++)
{
int j;
for (j = 0; j < M; j++)
{
if (t[mid] != p[j]) break;
if (p[j] < t[mid]) { end = mid - 1; }
else if (p[j] > t[mid]) { start = mid + 1; }
}
if (j == M)
{
printf("Pattern found at index %d \n", i);
}
}
}
if (start > end) cout << "Search has ended: pattern p does not occur in the text." << endl;
return 0;
}

Answer 1

Your algorithm is still a binary search. You split the array into two partitions, then select a partition, based on the value of a letter.
The requirements of a partitioned search is to have an ordered collection.
Let's use your example.
0 1 2 3 4 5 6 7 8 9 10 11
+---+---+---+---+---+---+---+---+---+---+---+---+
| a | b | c | d | e | a | b | b | d | f | e | g |
+---+---+---+---+---+---+---+---+---+---+---+---+
If you choose the midpoint at index 5, this yields the letter a. Since you are searching for the letter c first, then d, the algorithm says that the letter c must lie in the partition 6..11. Thus the basis of your issue.
The algorithm will not find cd because there is no c in the partition 6..11.
The algorithm assumes that the array is sorted and that for any given index, there will be one partition containing values less than array[index] and one partition containing values greater than array[index].
This assumption is demonstrated by the following code of yours:
if (p[j] < t[mid]) { end = mid - 1; }
else if (p[j] > t[mid]) { start = mid + 1; }
No matter how you name your algorithm, if it assumes an ordering on the array (e.g. p[j] < t[mid]), the array must be ordered.
You data is not ordered, so your algorithm fails the assumption, and thus the algorithm fails.
Edit 1:
Using Partitions
If you really must use a partitioning algorithm, you will need to build a set of partitions.
For example one partition starts at index 0 and proceeds until array[i] > array[i+1], this ends up at index 4. The other partition is 5..11.
(By the way, by determining the partitions, you have used more operations than a linear search.)
At this point, how do you know which partition to choose?
You don't. The letter c, that you are searching for, lies between a and e in the first partition; and a through g in the second partition. Pick a partition. If not found in the partition, you will have to search the other partition.
By performing a binary search on either partition, you have used more operations than a linear search.