|
| 1 | +#!/usr/bin/env python3 |
| 2 | + |
| 3 | +def build_suffix_array(s: str) -> list[int]: |
| 4 | + """ |
| 5 | + Build the suffix array for the given string. |
| 6 | +
|
| 7 | + Parameters: |
| 8 | + s (str): The input string. |
| 9 | +
|
| 10 | + Returns: |
| 11 | + list[int]: The suffix array (a list of starting indices of |
| 12 | + suffixes in sorted order). |
| 13 | + """ |
| 14 | + suffixes = [(s[i:], i) for i in range(len(s))] |
| 15 | + suffixes.sort() # Sort the suffixes lexicographically |
| 16 | + suffix_array = [suffix[1] for suffix in suffixes] |
| 17 | + return suffix_array |
| 18 | + |
| 19 | +def build_lcp_array(s: str, suffix_array: list[int]) -> list[int]: |
| 20 | + """ |
| 21 | + Build the LCP array for the given string and suffix array. |
| 22 | +
|
| 23 | + Parameters: |
| 24 | + s (str): The input string. |
| 25 | + suffix_array (list[int]): The suffix array. |
| 26 | +
|
| 27 | + Returns: |
| 28 | + list[int]: The LCP array. |
| 29 | + """ |
| 30 | + n = len(s) |
| 31 | + rank = [0] * n |
| 32 | + lcp = [0] * n |
| 33 | + |
| 34 | + # Compute the rank of each suffix |
| 35 | + for i, suffix_index in enumerate(suffix_array): |
| 36 | + rank[suffix_index] = i |
| 37 | + |
| 38 | + # Compute the LCP array |
| 39 | + h = 0 |
| 40 | + for i in range(n): |
| 41 | + if rank[i] > 0: |
| 42 | + j = suffix_array[rank[i] - 1] |
| 43 | + while (i + h < n) and (j + h < n) and (s[i + h] == s[j + h]): |
| 44 | + h += 1 |
| 45 | + lcp[rank[i]] = h |
| 46 | + if h > 0: |
| 47 | + h -= 1 # Decrease h for the next suffix |
| 48 | + return lcp |
| 49 | + |
| 50 | +# Example usage |
| 51 | +if __name__ == "__main__": |
| 52 | + s = "banana" |
| 53 | + suffix_array = build_suffix_array(s) |
| 54 | + lcp_array = build_lcp_array(s, suffix_array) |
| 55 | + |
| 56 | + print("Suffix Array:") |
| 57 | + for i in range(len(suffix_array)): |
| 58 | + print(f"{suffix_array[i]}: {s[suffix_array[i]:]}") |
| 59 | + |
| 60 | + print("\nLCP Array:") |
| 61 | + for i in range(1, len(lcp_array)): |
| 62 | + print(f"LCP between {s[suffix_array[i - 1]:]} and {s[suffix_array[i]]}: {lcp_array[i]}") |
0 commit comments