increase max hanzi length

src/include/dictdef.h

@@ -56,7 +56,10 @@ const size_t kHalfSpellingIdNum = 29;
 // stored.
 // -1 is because that 0 is never used.
 const size_t kMaxSpellingNum = 512 - kHalfSpellingIdNum - 1;
-const size_t kMaxSearchSteps = 40;
+// Increased to allow longer Pinyin input and longer sentence output.
+// Note: changing this requires adjusting related pools in matrixsearch.h
+// and dicttrie.h to avoid search failures under long input.
+const size_t kMaxSearchSteps = 128;
 
 // One character predicts its following characters.
 const size_t kMaxPredictSize = (kMaxLemmaSize - 1);

src/include/dicttrie.h

@@ -136,8 +136,10 @@ class DictTrie : AtomDictBase {
     bool save_dict(FILE *fp);
 #endif  // ___BUILD_MODEL___
 
-    static const int kMaxMileStone = 100;
-    static const int kMaxParsingMark = 600;
+    // Increase limits to support longer Pinyin input.
+    // If kMaxSearchSteps is increased, these must scale accordingly.
+    static const int kMaxMileStone = static_cast<int>(kMaxSearchSteps * 8);
+    static const int kMaxParsingMark = static_cast<int>(kMaxSearchSteps * 64);
     static const MileStoneHandle kFirstValidMileStoneHandle = 1;
 
     friend class DictParser;

src/include/matrixsearch.h

@@ -144,14 +144,16 @@ class MatrixSearch {
     static const size_t kMaxNodeARow = 5;
 
     // The maximum length of the sentence candidates counted in chinese
-    // characters
-    static const size_t kMaxSentenceLength = 16;
+    // characters.
+    static const size_t kMaxSentenceLength = 64;
 
     // The size of the matrix node pool.
-    static const size_t kMtrxNdPoolSize = 200;
+    // Scale with kMaxSearchSteps to avoid pool exhaustion for long input.
+    static const size_t kMtrxNdPoolSize = kMaxRowNum * kMaxNodeARow;
 
     // The size of the DMI node pool.
-    static const size_t kDmiPoolSize = 800;
+    // Scale with kMaxSearchSteps to avoid pool exhaustion for long input.
+    static const size_t kDmiPoolSize = kMaxRowNum * 20;
 
     // Used to indicate whether this object has been initialized.
     bool inited_;

tests/main.cpp

@@ -97,7 +97,8 @@ int main() {
 #ifdef _WIN32
     SetConsoleOutputCP(CP_UTF8);
 #endif
-    ime_pinyin::im_set_max_lens(64, 32);
+    // Keep output up to 64 Hanzi; input length up to 128 Pinyin chars.
+    ime_pinyin::im_set_max_lens(128, 64);
     if (!ime_pinyin::im_open_decoder("./data/dict_pinyin.dat", "./data/user_dict.dat")) {
         std::cout << "fany bug.\n";
         return 0;
@@ -119,5 +120,7 @@ int main() {
     test_pinyin_search_when_retriving_first_element("ni'shuo'ne'ma'de");
     test_pinyin_search_when_retriving_first_element("wei'ni'qian'gua'bu'bian");
     test_pinyin_search_when_retriving_first_element("wo'lai'kan'kan'zhe'ge'zui'duo'neng'shu'ru'duo'shao'ge'zi'kan'shang'qu'shi'bi'jiao'he'li'de");
+    test_pinyin_search_when_retriving_first_element("zheshikeyishuodemawojuedehaishiyaoweiwanyidianbijiaohaobanishuoneqishizhongguorenshizhegeyangzide");
+    test_pinyin_search_when_retriving_first_element("chulichangjuzishishenme");
     return 0;
-}
+}