RGBCube/serenity
1
Fork 0
mirror of https://github.com/RGBCube/serenity synced 2025-07-28 09:37:44 +00:00
Code Issues Activity Actions

AK: Implement memmem() for iterator haystacks

Browse source 
This uses the KMP algorithm to implement the search.
Also replaces the slow route of the normal memmem() with KMP, which
should be fairly faster (O(n + m) as opposed to O(n * m)) :^)
This commit is contained in:
AnotherTest 2020-12-05 04:29:09 +03:30 • committed by Andreas Kling
parent f12c98b29f
commit 036b39cdfd
2 changed files with 158 additions and 8 deletions
Download patch file Download diff file Expand all files Collapse all files

76
AK/MemMem.h
View file

@ -26,8 +26,11 @@
#pragma once
#include <AK/Array.h>
#include <AK/Assertions.h>
#include <AK/Span.h>
#include <AK/Types.h>
#include <AK/Vector.h>
namespace AK {
@ -59,6 +62,64 @@ const static void* bitap_bitwise(const void* haystack, size_t haystack_length, c
}
}
template<typename HaystackIterT>
static inline Optional<size_t> memmem(const HaystackIterT& haystack_begin, const HaystackIterT& haystack_end, Span<const u8> needle) requires(requires { (*haystack_begin).data(); (*haystack_begin).size(); })
{
auto prepare_kmp_partial_table = [&] {
Vector<int, 64> table;
table.resize(needle.size());
size_t position = 1;
int candidate = 0;
table[0] = -1;
while (position < needle.size()) {
if (needle[position] == needle[candidate]) {
table[position] = table[candidate];
} else {
table[position] = candidate;
do {
candidate = table[candidate];
} while (candidate >= 0 && needle[candidate] != needle[position]);
}
++position;
++candidate;
}
return table;
};
auto table = prepare_kmp_partial_table();
size_t total_haystack_index = 0;
size_t current_haystack_index = 0;
int needle_index = 0;
auto haystack_it = haystack_begin;
while (haystack_it != haystack_end) {
auto&& chunk = *haystack_it;
if (current_haystack_index >= chunk.size()) {
current_haystack_index = 0;
++haystack_it;
continue;
}
if (needle[needle_index] == chunk[current_haystack_index]) {
++needle_index;
++current_haystack_index;
++total_haystack_index;
if ((size_t)needle_index == needle.size())
return total_haystack_index - needle_index;
continue;
}
needle_index = table[needle_index];
if (needle_index < 0) {
++needle_index;
++current_haystack_index;
++total_haystack_index;
}
}
return {};
}
static inline const void* memmem(const void* haystack, size_t haystack_length, const void* needle, size_t needle_length)
{
if (needle_length == 0)
@ -73,15 +134,14 @@ static inline const void* memmem(const void* haystack, size_t haystack_length, c
if (needle_length < 32)
return bitap_bitwise(haystack, haystack_length, needle, needle_length);
// Fallback to a slower search.
auto length_diff = haystack_length - needle_length;
for (size_t i = 0; i < length_diff; ++i) {
const auto* start = ((const u8*)haystack) + i;
if (__builtin_memcmp(start, needle, needle_length) == 0)
return start;
}
// Fallback to KMP.
Array<Span<const u8>, 1> spans { Span<const u8> { (const u8*)haystack, haystack_length } };
auto result = memmem(spans.begin(), spans.end(), { (const u8*)needle, needle_length });
return nullptr;
if (result.has_value())
return (const u8*)haystack + result.value();
return {};
}
}