Add wildcard pattern matching algorithm using FFT

Author: OnePunchMonk

strings/wildcard_pattern_matching_fft.py

@@ -0,0 +1,96 @@
+import numpy as np
+from numpy.fft import fft, ifft
+
+def preprocess_text_and_pattern(text, pattern):
+    """Preprocesses text and pattern for pattern matching.
+
+    Args:
+        text: The input text string.
+        pattern: The input pattern string, potentially containing wildcards ('*').
+
+    Returns:
+        A tuple containing:
+            - A list of integers representing the text characters.
+            - A list of integers representing the pattern characters, with 0 for wildcards.
+    """
+
+    unique_chars = set(text + pattern)
+    char_to_int = {char: i + 1 for i, char in enumerate(unique_chars)}  # Unique non-zero integers
+
+    # Replace pattern '*' with 0, other characters with their unique integers
+    pattern_int = [char_to_int[char] if char != '*' else 0 for char in pattern]
+    text_int = [char_to_int[char] for char in text]
+
+    return text_int, pattern_int
+
+def fft_convolution(a, b):
+    """Performs convolution using the Fast Fourier Transform (FFT).
+
+    Args:
+        a: The first sequence.
+        b: The second sequence.
+
+    Returns:
+        The convolution of the two sequences.
+    """
+
+    n = len(a) + len(b) - 1
+    A = fft(a, n)
+    B = fft(b, n)
+    return np.real(ifft(A * B))
+
+def compute_A_fft(text_int, pattern_int):
+    """Computes the A array for the pattern matching algorithm.
+
+    Args:
+        text_int: The integer representation of the text.
+        pattern_int: The integer representation of the pattern.
+
+    Returns:
+        The A array.
+    """
+
+    n = len(text_int)
+    m = len(pattern_int)
+
+    # Power transforms of the pattern and text based on the formula
+    p1 = np.array(pattern_int)
+    p2 = np.array([p**2 for p in pattern_int])
+    p3 = np.array([p**3 for p in pattern_int])
+
+    t1 = np.array(text_int)
+    t2 = np.array([t**2 for t in text_int])
+    t3 = np.array([t**3 for t in text_int])
+
+    # Convolution to calculate the terms for A[i]
+    sum1 = fft_convolution(p3[::-1], t1)
+    sum2 = fft_convolution(p2[::-1], t2)
+    sum3 = fft_convolution(p1[::-1], t3)
+
+    # Calculate A[i] using the convolution results
+    A = sum1[:n - m + 1] - 2 * sum2[:n - m + 1] + sum3[:n - m + 1]
+
+    return A
+
+# Main function to run the matching
+if __name__ == "__main__":
+
+    import doctest
+    doctest.testmod()
+    # Get text and pattern as input from the user
+    # text = input("Enter the text: ")
+    # pattern = input("Enter the pattern (use '*' for wildcard): ")
+
+    text = "abcabc"
+    pattern = "abc*"
+
+
+
+
+    text_int, pattern_int = preprocess_text_and_pattern(text, pattern)
+    A = compute_A_fft(text_int, pattern_int)
+
+    # Matches occur where A[i] == 0
+    matches = [i for i in range(len(A)) if np.isclose(A[i], 0)]
+    print("Pattern matches at indices:", matches)
+