diff options
| author | mzuenni <mzuenni@users.noreply.github.com> | 2024-07-28 22:54:40 +0200 |
|---|---|---|
| committer | GitHub <noreply@github.com> | 2024-07-28 22:54:40 +0200 |
| commit | 8d11c6c8213f46f0fa19826917c255edd5d43cb1 (patch) | |
| tree | 96d75baff33d5a04b5a60f1a41f514a26c716874 /content/string/ahoCorasick.cpp | |
| parent | 8c33b4e0d3030cfed17fc64b4fe41133339f6d87 (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 'content/string/ahoCorasick.cpp')
| -rw-r--r-- | content/string/ahoCorasick.cpp | 52 |
1 files changed, 52 insertions, 0 deletions
diff --git a/content/string/ahoCorasick.cpp b/content/string/ahoCorasick.cpp new file mode 100644 index 0000000..eac312c --- /dev/null +++ b/content/string/ahoCorasick.cpp @@ -0,0 +1,52 @@ +constexpr ll ALPHABET_SIZE = 26, OFFSET = 'a'; +struct AhoCorasick { + struct vert { + int suffix = 0, ch, cnt = 0; + array<int, ALPHABET_SIZE> nxt = {}; + + vert(int p, int c) : suffix(-p), ch(c) {} + }; + vector<vert> aho = {{0, -1}}; + + int addString(string &s) { + int v = 0; + for (auto c : s) { + int idx = c - OFFSET; + if (!aho[v].nxt[idx]) { + aho[v].nxt[idx] = sz(aho); + aho.emplace_back(v, idx); + } + v = aho[v].nxt[idx]; + } + aho[v].cnt++; + return v; // trie node index of pattern (pattern state) + } + + int getSuffix(int v) { + if (aho[v].suffix < 0) { + aho[v].suffix = go(getSuffix(-aho[v].suffix), aho[v].ch); + } + return aho[v].suffix; + } + + int go(int v, int idx) { // Root is v=0, idx is char - OFFSET + if (aho[v].nxt[idx]) return aho[v].nxt[idx]; + else return v == 0 ? 0 : go(getSuffix(v), idx); + } + + vector<vector<int>> adj; + vector<ll> dp; + void buildGraph() { + adj.resize(sz(aho)); + dp.assign(sz(aho), 0); + for (int i = 1; i < sz(aho); i++) { + adj[getSuffix(i)].push_back(i); + }} + + void dfs(int v = 0) { // dp on tree + for (int u : adj[v]) { + //dp[u] = dp[v] + aho[u].cnt; // pattern count + dfs(u); + dp[v] += dp[u]; // no of matches + }} +}; |