summaryrefslogtreecommitdiff
path: root/test/string/ahoCorasick.cpp
diff options
context:
space:
mode:
authormzuenni <mzuenni@users.noreply.github.com>2024-07-28 22:54:40 +0200
committerGitHub <noreply@github.com>2024-07-28 22:54:40 +0200
commit8d11c6c8213f46f0fa19826917c255edd5d43cb1 (patch)
tree96d75baff33d5a04b5a60f1a41f514a26c716874 /test/string/ahoCorasick.cpp
parent8c33b4e0d3030cfed17fc64b4fe41133339f6d87 (diff)
Test (#4)
* update * moved content in subdir * rename file * add test setup * add test setup * add github action * automaticly test all cpp files * timeout after 10s * setulimit and dont zero memory * test build pdf * install latexmk * update * update * ngerman * fonts * removed old code * add first test * added tests * test in sorted order * more tests * simplified test * more tests * fix suffix tree * fixes and improvements * done ust lst directly * fix swap * add links to pdf * fix constants * add primorial * add comment * various improvements * more tests * added missing stuf * more tests * fix tests * more tests * more tests * more tests * fix recursion? * test trie * more tests * only use python temporarily for listings * only use python temporarily for listings * more tests * fix longestCommonSubstring * more tests * more tests * made code more similiar * fix? * more tests * more tests * more tests * add ahoCorasick test + limit 4GB stack size * more tests * fix test * add additional test * more tests * more tests * fix? * better fix * fix virtual tree * more tests * more tests * recursive closest pair * more tests * decrease limit * new tests * more tests * fix name * more tests * add test * new test * more tests * more tests * more tests * more tests * new test and content * new code * new code * larger tests * fix and test * new test * new test * update pdf * remove comments * new test * more tests * more testcases * more tests * increased limit * more tests * more tests * more tests * new tests * more tests * shortened code * new test * add basic tests for bigint * more tests * removed old files * new test * ignore some files * more auto more ccw * fix test * more tests * fix * new tests * more tests * more tests * stronger test * actually verify delaunay... * more tests * fix header * more tests * run tests parallel? * test parralel? * add --missing * separate workflows * test * is the pdf checked? * separate workflows * fix workflow * more workflows --------- Co-authored-by: Yidi <noob999noob999@gmail.com>
Diffstat (limited to 'test/string/ahoCorasick.cpp')
-rw-r--r--test/string/ahoCorasick.cpp76
1 files changed, 76 insertions, 0 deletions
diff --git a/test/string/ahoCorasick.cpp b/test/string/ahoCorasick.cpp
new file mode 100644
index 0000000..c3361d6
--- /dev/null
+++ b/test/string/ahoCorasick.cpp
@@ -0,0 +1,76 @@
+#include "../util.h"
+#include <string/ahoCorasick.cpp>
+
+vector<ll> naive(string s, vector<string> patterns) {
+ vector<ll> ans(patterns.size());
+ for (int k = 0; k < (int)patterns.size(); k++) {
+ string pattern = patterns[k];
+ for (int i = 0; i + pattern.size() <= s.size(); i++) {
+ if (s.substr(i, pattern.size()) == pattern) ans[k]++;
+ }
+ }
+ return ans;
+}
+
+vector<ll> normal(string s, vector<string> patterns) {
+ AhoCorasick aho;
+ vector<int> ind(patterns.size());
+ for (int i = 0; i < (int)patterns.size(); i++) {
+ ind[i] = aho.addString(patterns[i]);
+ }
+ aho.buildGraph();
+
+ int v = 0;
+ for (char c : s) v = aho.go(v, c - OFFSET), aho.dp[v]++;
+ aho.dfs();
+ vector<ll> ans(patterns.size());
+ for (int i = 0; i < (int)patterns.size(); i++) {
+ ans[i] = aho.dp[ind[i]];
+ }
+ return ans;
+}
+
+void stress_test() {
+ ll queries = 0;
+ for (int i = 0; i < 100; i++) {
+ int n = Random::integer(1, 100);
+ string s = Random::string(n, "abc");
+ int m = Random::integer(1, 100);
+ vector<string> patterns(m);
+ for (string& e : patterns) {
+ int k = Random::integer(1, 100);
+ e = Random::string(k, "abc");
+ }
+
+ auto got = normal(s, patterns);
+ auto expected = naive(s, patterns);
+ if (got != expected) cerr << "Wrong Answer" << FAIL;
+ queries++;
+ }
+ cerr << "Tested random queries: " << queries << endl;
+}
+
+constexpr int N = 1'000'000;
+void performance_test() {
+ timer t;
+ string s = string(N, 'a') + Random::string(N, "ab");
+ vector<string> patterns = {"a"};
+ for (int sm = 1; sm < N; sm += patterns.back().size()) {
+ patterns.emplace_back(patterns.back().size()+1, 'a');
+ }
+ for (int i = 0; i < 100; i++) {
+ patterns.emplace_back(Random::string(N/100, "ab"));
+ }
+
+ t.start();
+ hash_t hash = normal(s, patterns)[0];
+ t.stop();
+
+ if (t.time > 500) cerr << "Too slow: " << t.time << FAIL;
+ cerr << "Tested performance: " << t.time << "ms (hash: hash " << hash << ")" << endl;
+}
+
+int main() {
+ stress_test();
+ performance_test();
+}