King Fahd University of Petroleum and Minerals

College of Computing and Mathematics

Information & Computer Science Department

ICS 202
Data Structures and Algorithms

Unit 11
String Matching
Reading Assignment

›
◼
◼
Outline


Introduction


›
›
›
›
›
›
›
Straightforward (Brute Force) Algorithm

›
›
›
›
 6
Brute Force algorithm...


›
›

The Knuth-Morris-Pratt Algorithm








KMP Algorithm - Motivation

✓
✓ ✓
✓ ✓ ✓
✓ ✓ ✓ ✓
✓ ✓ ✓ ✓ 

›
›

i=4
j=2
KMP Algorithm - Mechanism


KMP Algorithm – Preprocessing Terms

 0. . 𝑛 − 1 0. . 𝑗 𝑗
𝑛−2
◼ 𝜆

 0. . 𝑛 − 1 𝑗. . 𝑛 − 1 𝑗
𝑛−1
◼ 𝜆

 0. . 𝑛 − 1

 𝜆

› 𝜆
› 𝜆
›
KMP Algorithm - Preprocessing

 l 𝜋−


l

If the border is 𝜆
KMP Algorithm – The lps- or next-array


𝒋
P[j]
𝑃[0. . 𝑗 − 1]

𝑛𝑒𝑥𝑡[𝑗]

𝒋
P[j]
𝑃[0. . 𝑗 − 1]

𝜆 𝜆 𝜆 𝜆 𝜆 𝜆

𝜆 𝜆 𝜆 𝜆 𝜆 𝜆

𝑛𝑒𝑥𝑡[𝑗]


The KMP Algorithm – Principle Explained






KMP Algorithm Search Methodology

Let T be the text

Let P be the pattern.
If there is a mismatch at index j > 0 of P
P is shifted j – next[j] to the right of T;
Restart searching P at index next[j];
Else // there is a mismatch at index j == 0
P is shifted 1 to the right of text T; // same as j – next[j] == 0 – (-1)
Restart searching the pattern at index 0.
The KMP Algorithm – Tracing example 01

𝜆 𝜆 𝜆 𝜆

𝜆 𝜆 𝜆 𝜆

𝑛𝑒𝑥𝑡[𝑗]

›
›
›
i

j
The KMP Algorithm – Tracing example 01...


›
i

 →
The KMP Algorithm – Tracing example 01...


›
i

 →
The KMP Algorithm – Tracing example 01...


›
i

 →
The KMP Algorithm – Tracing example 01...


›
i

 →
The KMP Algorithm – Tracing example 01...


›
i

 →
The KMP Algorithm – Tracing example 01...


›
i

j
›
›
The KMP Algorithm


The KMP Algorithm
public static String searchKMP(String pattern, String text) static int[] computeLPSArray(String pattern) {
{
int M = pattern.length(); int M = pattern.length();
int N = text.length(); int lps[] = new int[M];
String indexes = "";
// length of the previous longest prefix suffix
// Preprocess the pattern (calculate lps[] array) int j = 0;
// lps[] will hold the longest prefix suffix values
for pattern int i = 1;
int[] lps = computeLPSArray(pattern);
lps[0] = 0; // lps[0] is always 0
int i = 0; // index for txt[]
int j = 0; // index for pat[]
while (i < N) { // the loop calculates lps[i] for i = 1 to M-1
if (pattern.charAt(j) == text.charAt(i)) {
j++; while (i < M) {
i++; if (pattern.charAt(i) == pattern.charAt(j)) {
}
if (j == M) { j++;
indexes += (i - j) + " ";
j = lps[j - 1]; lps[i] = j;
} i++;
// mismatch after j matches }
else if (i < N && pattern.charAt(j) !=
text.charAt(i)) { else // (pat[i] != pat[len])
// Do not match lps[0..lps[j-1]] characters, {
they will match anyway
if (j != 0) if (j != 0) {
j = lps[j - 1];
else j = lps[j - 1];
i = i + 1; }
}
} else // if (len == 0)
return indexes; {
} lps[i] = j;
i++;
return lps; }
}
}
}
The KMP Algorithm – Tracing example 02


Comparison of Brute Force and KMP Algorithms


 
Some applications of String matching
Exercises


›


›
Exercises...