From: Michael Tremer <michael.tremer@ipfire.org>
Date: Sat, 4 Jan 2025 15:21:03 +0000 (+0000)
Subject: strings: Implement a fast search that even works over large buffers
X-Git-Tag: 0.9.30~539
X-Git-Url: http://git.ipfire.org/cgi-bin/gitweb.cgi?a=commitdiff_plain;h=33d627d1d7244f4bef9f8b1210c11120538fe24d;p=pakfire.git

strings: Implement a fast search that even works over large buffers

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>
---

diff --git a/src/pakfire/string.c b/src/pakfire/string.c
index e409bba3c..7e59d09af 100644
--- a/src/pakfire/string.c
+++ b/src/pakfire/string.c
@@ -309,6 +309,80 @@ char* pakfire_string_join(const char** list, const char* delim) {
 	return string;
 }
 
+int pakfire_string_search(const char* haystack, ssize_t lhaystack,
+		const char* needle, ssize_t lneedle) {
+	// Check inputs
+	if (!haystack || !needle || !lhaystack || !lneedle)
+		return -EINVAL;
+
+	// Fill lengths
+	if (lhaystack < 0)
+		lhaystack = strlen(haystack);
+
+	if (lneedle < 0)
+		lneedle = strlen(needle);
+
+	// Alphabet size
+	const int d = 256;
+
+	// A prime number
+	const int q = 101;
+
+	// Hash constant
+	int c = 1;
+
+	// Hash value for the haystack
+	int h = 0;
+
+	// Hash value for the needle
+	int n = 0;
+
+	// Compute the hash constant
+	for (int i = 0; i < lneedle - 1; i++)
+		c = (c * d) % q;
+
+	// Calculate the hash value of pattern and first window
+	for (int i = 0; i < lneedle; i++) {
+		h = (d * h + haystack[i]) % q;
+		n = (d * n + needle[i]) % q;
+    }
+
+	// Slide the text over the pattern...
+	for (int i = 0; i <= lhaystack - lneedle; i++) {
+		// Check the hash values of the current window.
+		// If they match, then check then check if we actually found a match
+		if (h == n) {
+			// Assume there was a match
+			int match = 1;
+
+			for (int j = 0; j < lneedle; j++) {
+				// Skip the parts that match
+				if (haystack[i + j] == needle[j])
+					continue;
+
+				// If a character does not match, we reset and continue
+				match = 0;
+				break;
+			}
+
+			// We found a match and are done
+			if (match)
+				return i;
+		}
+
+		// Calculate the hash value for the next window
+		if (i < lhaystack - lneedle) {
+			h = (d * (h - haystack[i] * c) + haystack[i + lneedle]) % q;
+
+			// If we have underflowed, we make the hash positive again
+			if (h < 0)
+				h += q;
+		}
+	}
+
+	return -1;
+}
+
 size_t pakfire_strings_length(char** array) {
 	size_t length = 0;
 
diff --git a/src/pakfire/string.h b/src/pakfire/string.h
index f9f0c2d81..42d72f00a 100644
--- a/src/pakfire/string.h
+++ b/src/pakfire/string.h
@@ -82,6 +82,9 @@ static inline int pakfire_string_contains_whitespace(const char* s) {
 	return 0;
 }
 
+int pakfire_string_search(const char* haystack, ssize_t lhaystack,
+	const char* needle, ssize_t lneedle);
+
 /*
 	String Arrays
 */
diff --git a/tests/libpakfire/string.c b/tests/libpakfire/string.c
index 2668336a0..75de6882d 100644
--- a/tests/libpakfire/string.c
+++ b/tests/libpakfire/string.c
@@ -347,6 +347,27 @@ FAIL:
 	return r;
 }
 
+static int test_string_search(const struct test* t) {
+	int r = EXIT_FAILURE;
+
+	// Check if we find the pattern in the right place
+	ASSERT_EQUALS(pakfire_string_search("ABCDEFGHIJKLMNOPQ", -1, "ABC", -1), 0);
+	ASSERT_EQUALS(pakfire_string_search("ABCDEFGHIJKLMNOPQ", -1, "DEF", -1), 3);
+	ASSERT_EQUALS(pakfire_string_search("ABCDEFGHIJKLMNOPQ", -1, "OPQ", -1), 14);
+
+	// Check if don't find some false positive
+	ASSERT_EQUALS(pakfire_string_search("ABCDEFGHIJKLMNOPQ", -1, "XYZ", -1), -1);
+
+	// Check invalid inputs
+	ASSERT_ERROR(pakfire_string_search(NULL, -1, NULL, -1), EINVAL);
+
+	// Everything passed
+	r = EXIT_SUCCESS;
+
+FAIL:
+	return r;
+}
+
 int main(int argc, const char* argv[]) {
 	testsuite_add_test(test_string_set, 0);
 	testsuite_add_test(test_string_startswith, 0);
@@ -359,6 +380,7 @@ int main(int argc, const char* argv[]) {
 	testsuite_add_test(test_parse_bytes, 0);
 	testsuite_add_test(test_intervals, 0);
 	testsuite_add_test(test_string_contains_whitespace, 0);
+	testsuite_add_test(test_string_search, 0);
 
 	return testsuite_run(argc, argv);
 }