1 files changed, 114 insertions, 0 deletions

@@ -0,0 +1,114 @@
+#ifndef GENERATOR_H_D5C6A724
+#define GENERATOR_H_D5C6A724
+
+#include <optional>
+#include <set>
+#include <string>
+#include <unordered_map>
+#include <vector>
+
+enum PuzzleType {
+  kWhiteTop = 0,
+  kWhiteBottom = 1,
+  kYellowTop = 2,
+  kYellowMiddle = 3,
+  kBlackTop = 4,
+  kBlackMiddle = 5,
+};
+
+class generator {
+ public:
+  // Constructor
+
+  generator(std::string agidPath, std::string wordNetPath,
+            std::string cmudictPath, std::string wordfreqPath,
+            std::string outputPath);
+
+  // Action
+
+  void run();
+
+ private:
+  // Helpers
+
+  size_t LookupOrCreatePronunciation(const std::string& phonemes);
+
+  size_t LookupOrCreateForm(const std::string& form);
+
+  size_t LookupOrCreateWord(const std::string& word);
+
+  void AddPronunciationToForm(size_t pronunciation_id, size_t form_id);
+
+  void AddFormToWord(size_t form_id, size_t word_id);
+
+  void AddWordToSynset(size_t word_id, int wnid);
+
+  void AddFormToAnagramSet(size_t form_id, const std::string& sorted_letters);
+
+  void AddPronunciationToAnaphoneSet(size_t pronunciation_id,
+                                     const std::string& sorted_phonemes);
+
+  // Input
+
+  std::string agidPath_;
+  std::string wordNetPath_;
+  std::string cmudictPath_;
+  std::string wordfreqPath_;
+
+  // Output
+
+  std::string outputPath_;
+
+  // Indexes
+
+  struct Pronunciation {
+    size_t id;
+    std::string phonemes;
+    std::string prerhyme;
+    std::string rhyme;
+    std::vector<size_t> form_ids;
+    std::optional<size_t> anaphone_set_id;
+    std::string stressless_phonemes;
+  };
+
+  struct Form {
+    size_t id;
+    std::string text;
+    std::vector<size_t> word_ids;
+    std::vector<size_t> pronunciation_ids;
+    std::optional<size_t> anagram_set_id;
+    std::optional<size_t> reverse_form_id;
+
+    std::unordered_map<PuzzleType, std::set<size_t>> puzzles;
+  };
+
+  struct Word {
+    size_t id;
+    size_t base_form_id;
+    std::vector<size_t> form_ids;
+    std::vector<size_t> synsets;
+  };
+
+  std::vector<Pronunciation> pronunciations_;
+  std::unordered_map<std::string, size_t> pronunciation_by_phonemes_;
+  std::unordered_map<std::string, std::vector<size_t>> pronunciations_by_rhyme_;
+  std::unordered_map<std::string, std::vector<size_t>>
+      pronunciations_by_blank_phonemes_;
+
+  std::vector<std::vector<size_t>> anaphone_sets_;
+  std::unordered_map<std::string, size_t> anaphone_set_by_sorted_phonemes_;
+
+  std::vector<Form> forms_;
+  std::unordered_map<std::string, size_t> form_by_text_;
+
+  std::vector<std::vector<size_t>> anagram_sets_;
+  std::unordered_map<std::string, size_t> anagram_set_by_sorted_letters_;
+
+  std::vector<Word> words_;
+  std::unordered_map<std::string, size_t> word_by_base_;
+
+  std::vector<std::vector<size_t>> synsets_;
+  std::unordered_map<int, size_t> synset_by_wnid_;
+};
+
+#endif /* end of include guard: GENERATOR_H_D5C6A724 */
\ No newline at end of file