opencc-wasm 0.8.2 → 0.10.0

package/dist/data/config/opencc_config.schema.json

@@ -3,151 +3,102 @@
   "id": "https://opencc.byvoid.com/schema/opencc_config.schema.json",
   "title": "OpenCC configuration",
   "type": "object",
-  "required": [
-    "segmentation",
-    "conversion_chain"
-  ],
+  "required": ["name", "conversion_chain"],
   "additionalProperties": false,
   "properties": {
-    "name": {
-      "type": "string"
-    },
-    "segmentation": {
-      "$ref": "#/definitions/segmentation"
+    "name": { "type": "string" },
+    "normalization": {
+      "type": "array",
+      "minItems": 1,
+      "items": { "$ref": "#/definitions/conversion" }
     },
+    "segmentation": { "$ref": "#/definitions/segmentation" },
     "conversion_chain": {
       "type": "array",
       "minItems": 1,
-      "items": {
-        "$ref": "#/definitions/conversion"
-      }
+      "items": { "$ref": "#/definitions/conversion" }
     }
   },
   "definitions": {
     "segmentation": {
       "anyOf": [
-        {
-          "$ref": "#/definitions/mmseg_segmentation"
-        },
-        {
-          "$ref": "#/definitions/plugin_segmentation"
-        }
+        { "$ref": "#/definitions/mmseg_segmentation" },
+        { "$ref": "#/definitions/plugin_segmentation" }
       ]
     },
     "mmseg_segmentation": {
       "type": "object",
-      "required": [
-        "type",
-        "dict"
-      ],
+      "required": ["type", "dict"],
       "additionalProperties": false,
       "properties": {
-        "type": {
-          "enum": [
-            "mmseg"
-          ]
-        },
-        "dict": {
-          "$ref": "#/definitions/dict"
-        }
+        "type": { "enum": ["mmseg"] },
+        "dict": { "$ref": "#/definitions/dict" }
       }
     },
     "plugin_segmentation": {
       "type": "object",
-      "required": [
-        "type"
-      ],
+      "required": ["type"],
       "not": {
-        "properties": {
-          "type": {
-            "enum": [
-              "mmseg"
-            ]
-          }
-        },
-        "required": [
-          "type"
-        ]
+        "properties": { "type": { "enum": ["mmseg"] } },
+        "required": ["type"]
       },
       "properties": {
-        "type": {
-          "type": "string",
-          "minLength": 1
-        },
+        "type": { "type": "string", "minLength": 1 },
         "resources": {
           "type": "object",
-          "additionalProperties": {
-            "type": "string"
-          }
+          "additionalProperties": { "type": "string" }
         }
       },
-      "additionalProperties": {
-        "type": "string"
-      }
+      "additionalProperties": { "type": "string" }
     },
     "conversion": {
       "type": "object",
-      "required": [
-        "dict"
-      ],
+      "required": ["dict"],
       "additionalProperties": false,
-      "properties": {
-        "dict": {
-          "$ref": "#/definitions/dict"
-        }
-      }
+      "properties": { "dict": { "$ref": "#/definitions/dict" } }
     },
     "dict": {
       "anyOf": [
-        {
-          "$ref": "#/definitions/file_dict"
-        },
-        {
-          "$ref": "#/definitions/group_dict"
-        }
+        { "$ref": "#/definitions/file_dict" },
+        { "$ref": "#/definitions/inline_dict" },
+        { "$ref": "#/definitions/group_dict" }
       ]
     },
     "file_dict": {
       "type": "object",
-      "required": [
-        "type",
-        "file"
-      ],
+      "required": ["type", "file"],
       "additionalProperties": false,
       "properties": {
-        "type": {
-          "enum": [
-            "text",
-            "ocd",
-            "ocd2"
-          ]
-        },
-        "file": {
-          "type": "string",
-          "minLength": 1
+        "type": { "enum": ["text", "ocd", "ocd2"] },
+        "file": { "type": "string", "minLength": 1 },
+        "may_output_tofu": { "type": "boolean" }
+      }
+    },
+    "inline_dict": {
+      "type": "object",
+      "required": ["type", "entries"],
+      "additionalProperties": false,
+      "properties": {
+        "type": { "enum": ["inline"] },
+        "entries": {
+          "type": "object",
+          "additionalProperties": { "type": "string", "minLength": 1 }
         }
       }
     },
     "group_dict": {
       "type": "object",
-      "required": [
-        "type",
-        "dicts"
-      ],
+      "required": ["type", "dicts", "match_policy"],
       "additionalProperties": false,
       "properties": {
-        "type": {
-          "enum": [
-            "group"
-          ]
-        },
+        "type": { "enum": ["group"] },
+        "match_policy": { "enum": ["short_circuit", "union"] },
         "dicts": {
           "type": "array",
           "minItems": 1,
-          "items": {
-            "$ref": "#/definitions/dict"
-          }
-        }
+          "items": { "$ref": "#/definitions/dict" }
+        },
+        "may_output_tofu": { "type": "boolean" }
       }
     }
   }

package/dist/data/config/s2hk.json

@@ -1,27 +1,46 @@
 {
   "name": "Simplified Chinese to Traditional Chinese (Hong Kong variant)",
+  "normalization": [
+    { "dict": { "type": "ocd2", "file": "CJK_Compatibility_Ideographs.ocd2" } }
+  ],
   "segmentation": {
     "type": "mmseg",
-    "dict": {
-      "type": "ocd2",
-      "file": "STPhrases.ocd2"
-    }
-  },
-  "conversion_chain": [{
     "dict": {
       "type": "group",
-      "dicts": [{
-        "type": "ocd2",
-        "file": "STPhrases.ocd2"
-      }, {
-        "type": "ocd2",
-        "file": "STCharacters.ocd2"
-      }]
+      "match_policy": "union",
+      "dicts": [
+        { "type": "ocd2", "file": "STPhrases.ocd2" },
+        { "type": "ocd2", "file": "STPhrases_GeneratedFromRegionalPhrases.ocd2" }
+      ]
     }
-  }, {
-    "dict": {
-      "type": "ocd2",
-      "file": "HKVariants.ocd2"
+  },
+  "conversion_chain": [
+    {
+      "dict": {
+        "type": "group",
+        "match_policy": "short_circuit",
+        "dicts": [
+          {
+            "type": "group",
+            "match_policy": "union",
+            "dicts": [
+              { "type": "ocd2", "file": "STPhrases.ocd2" },
+              { "type": "ocd2", "file": "STPhrases_GeneratedFromRegionalPhrases.ocd2" }
+            ]
+          },
+          { "type": "ocd2", "file": "STCharacters.ocd2" }
+        ]
+      }
+    },
+    {
+      "dict": {
+        "type": "group",
+        "match_policy": "short_circuit",
+        "dicts": [
+          { "type": "ocd2", "file": "HKVariantsPhrases.ocd2" },
+          { "type": "ocd2", "file": "HKVariants.ocd2" }
+        ]
+      }
     }
-  }]
+  ]
 }

package/dist/data/config/s2hk_jieba.json

@@ -1,5 +1,13 @@
 {
   "name": "Simplified Chinese to Traditional Chinese (Hong Kong variant, Jieba Segmentation - Experimental)",
+  "normalization": [
+    {
+      "dict": {
+        "type": "ocd2",
+        "file": "CJK_Compatibility_Ideographs.ocd2"
+      }
+    }
+  ],
   "segmentation": {
     "type": "jieba",
     "resources": {
@@ -7,21 +15,48 @@
       "model_path": "jieba_dict/hmm_model.utf8"
     }
   },
-  "conversion_chain": [{
-    "dict": {
-      "type": "group",
-      "dicts": [{
-        "type": "ocd2",
-        "file": "STPhrases.ocd2"
-      }, {
-        "type": "ocd2",
-        "file": "STCharacters.ocd2"
-      }]
-    }
-  }, {
-    "dict": {
-      "type": "ocd2",
-      "file": "HKVariants.ocd2"
+  "conversion_chain": [
+    {
+      "dict": {
+        "type": "group",
+        "match_policy": "short_circuit",
+        "dicts": [
+          {
+            "type": "group",
+            "match_policy": "union",
+            "dicts": [
+              {
+                "type": "ocd2",
+                "file": "STPhrases.ocd2"
+              },
+              {
+                "type": "ocd2",
+                "file": "STPhrases_GeneratedFromRegionalPhrases.ocd2"
+              }
+            ]
+          },
+          {
+            "type": "ocd2",
+            "file": "STCharacters.ocd2"
+          }
+        ]
+      }
+    },
+    {
+      "dict": {
+        "type": "group",
+        "match_policy": "short_circuit",
+        "dicts": [
+          {
+            "type": "ocd2",
+            "file": "HKVariantsPhrases.ocd2"
+          },
+          {
+            "type": "ocd2",
+            "file": "HKVariants.ocd2"
+          }
+        ]
+      }
     }
-  }]
+  ]
 }

package/dist/data/config/s2hkp.json

@@ -0,0 +1,47 @@
+{
+  "name": "Simplified Chinese to Traditional Chinese (Hong Kong variant, with phrases)",
+  "normalization": [
+    { "dict": { "type": "ocd2", "file": "CJK_Compatibility_Ideographs.ocd2" } }
+  ],
+  "segmentation": {
+    "type": "mmseg",
+    "dict": {
+      "type": "group",
+      "match_policy": "union",
+      "dicts": [
+        { "type": "ocd2", "file": "STPhrases.ocd2" },
+        { "type": "ocd2", "file": "STPhrases_GeneratedFromRegionalPhrases.ocd2" }
+      ]
+    }
+  },
+  "conversion_chain": [
+    {
+      "dict": {
+        "type": "group",
+        "match_policy": "short_circuit",
+        "dicts": [
+          {
+            "type": "group",
+            "match_policy": "union",
+            "dicts": [
+              { "type": "ocd2", "file": "STPhrases.ocd2" },
+              { "type": "ocd2", "file": "STPhrases_GeneratedFromRegionalPhrases.ocd2" }
+            ]
+          },
+          { "type": "ocd2", "file": "STCharacters.ocd2" }
+        ]
+      }
+    },
+    {
+      "dict": {
+        "type": "group",
+        "match_policy": "short_circuit",
+        "dicts": [
+          { "type": "ocd2", "file": "HKPhrases.ocd2" },
+          { "type": "ocd2", "file": "HKVariantsPhrases.ocd2" },
+          { "type": "ocd2", "file": "HKVariants.ocd2" }
+        ]
+      }
+    }
+  ]
+}

package/dist/data/config/s2hkp_jieba.json

@@ -0,0 +1,66 @@
+{
+  "name": "Simplified Chinese to Traditional Chinese (Hong Kong variant, with phrases, Jieba Segmentation - Experimental)",
+  "normalization": [
+    {
+      "dict": {
+        "type": "ocd2",
+        "file": "CJK_Compatibility_Ideographs.ocd2"
+      }
+    }
+  ],
+  "segmentation": {
+    "type": "jieba",
+    "resources": {
+      "dict_path": "jieba_dict/jieba_merged.ocd2",
+      "model_path": "jieba_dict/hmm_model.utf8"
+    }
+  },
+  "conversion_chain": [
+    {
+      "dict": {
+        "type": "group",
+        "match_policy": "short_circuit",
+        "dicts": [
+          {
+            "type": "group",
+            "match_policy": "union",
+            "dicts": [
+              {
+                "type": "ocd2",
+                "file": "STPhrases.ocd2"
+              },
+              {
+                "type": "ocd2",
+                "file": "STPhrases_GeneratedFromRegionalPhrases.ocd2"
+              }
+            ]
+          },
+          {
+            "type": "ocd2",
+            "file": "STCharacters.ocd2"
+          }
+        ]
+      }
+    },
+    {
+      "dict": {
+        "type": "group",
+        "match_policy": "short_circuit",
+        "dicts": [
+          {
+            "type": "ocd2",
+            "file": "HKPhrases.ocd2"
+          },
+          {
+            "type": "ocd2",
+            "file": "HKVariantsPhrases.ocd2"
+          },
+          {
+            "type": "ocd2",
+            "file": "HKVariants.ocd2"
+          }
+        ]
+      }
+    }
+  ]
+}

package/dist/data/config/s2t.json

@@ -1,22 +1,25 @@
 {
-  "name": "Simplified Chinese to Traditional Chinese",
-  "segmentation": {
-    "type": "mmseg",
-    "dict": {
-      "type": "ocd2",
-      "file": "STPhrases.ocd2"
+  "name": "Simplified Chinese to Traditional Chinese (OpenCC Standard)",
+  "normalization": [
+    { "dict": { "type": "ocd2", "file": "CJK_Compatibility_Ideographs.ocd2" } }
+  ],
+  "conversion_chain": [
+    {
+      "dict": {
+        "type": "group",
+        "match_policy": "short_circuit",
+        "dicts": [
+          {
+            "type": "group",
+            "match_policy": "union",
+            "dicts": [
+              { "type": "ocd2", "file": "STPhrases.ocd2" },
+              { "type": "ocd2", "file": "STPhrases_GeneratedFromRegionalPhrases.ocd2" }
+            ]
+          },
+          { "type": "ocd2", "file": "STCharacters.ocd2" }
+        ]
+      }
     }
-  },
-  "conversion_chain": [{
-    "dict": {
-      "type": "group",
-      "dicts": [{
-        "type": "ocd2",
-        "file": "STPhrases.ocd2"
-      }, {
-        "type": "ocd2",
-        "file": "STCharacters.ocd2"
-      }]
-    }
-  }]
+  ]
 }

package/dist/data/config/s2t_cngov.json

@@ -1,12 +1,5 @@
 {
   "name": "Simplified Chinese to Traditional Chinese (The Chinese Government Standard)",
-  "author": "TerryTian-tech",
-  "license": "Apache License 2.0",
-  "source": "https://github.com/TerryTian-tech/OpenCC-Traditional-Chinese-characters-according-to-Chinese-government-standards",
-  "contributors": ["TerryTian-tech", "Yi Jianpeng", "Hu Xinmei", "Duan Yatong"],
-  "reference": "《通用规范汉字表》(2013)",
-  "description": "Conservative conversion that preserves intentional simplified characters in mixed documents while standardizing traditional characters only.",
-
   "segmentation": {
     "type": "mmseg",
     "dict": {
@@ -14,16 +7,21 @@
       "file": "cngov/STPhrases.ocd2"
     }
   },
-  "conversion_chain": [{
-    "dict": {
-      "type": "group",
-      "dicts": [{
-        "type": "ocd2",
-        "file": "cngov/STPhrases.ocd2"
-      }, {
-        "type": "ocd2",
-        "file": "cngov/STCharacters.ocd2"
-      }]
+  "conversion_chain": [
+    {
+      "dict": {
+        "type": "group",
+        "dicts": [
+          {
+            "type": "ocd2",
+            "file": "cngov/STPhrases.ocd2"
+          },
+          {
+            "type": "ocd2",
+            "file": "cngov/STCharacters.ocd2"
+          }
+        ]
+      }
     }
-  }]
+  ]
 }

package/dist/data/config/s2t_jieba.json

@@ -1,5 +1,13 @@
 {
   "name": "Simplified Chinese to Traditional Chinese (Jieba Segmentation - Experimental)",
+  "normalization": [
+    {
+      "dict": {
+        "type": "ocd2",
+        "file": "CJK_Compatibility_Ideographs.ocd2"
+      }
+    }
+  ],
   "segmentation": {
     "type": "jieba",
     "resources": {
@@ -7,16 +15,32 @@
       "model_path": "jieba_dict/hmm_model.utf8"
     }
   },
-  "conversion_chain": [{
-    "dict": {
-      "type": "group",
-      "dicts": [{
-        "type": "ocd2",
-        "file": "STPhrases.ocd2"
-      }, {
-        "type": "ocd2",
-        "file": "STCharacters.ocd2"
-      }]
+  "conversion_chain": [
+    {
+      "dict": {
+        "type": "group",
+        "match_policy": "short_circuit",
+        "dicts": [
+          {
+            "type": "group",
+            "match_policy": "union",
+            "dicts": [
+              {
+                "type": "ocd2",
+                "file": "STPhrases.ocd2"
+              },
+              {
+                "type": "ocd2",
+                "file": "STPhrases_GeneratedFromRegionalPhrases.ocd2"
+              }
+            ]
+          },
+          {
+            "type": "ocd2",
+            "file": "STCharacters.ocd2"
+          }
+        ]
+      }
     }
-  }]
+  ]
 }

package/dist/data/config/s2tw.json

@@ -1,27 +1,46 @@
 {
-  "name": "Simplified Chinese to Traditional Chinese (Taiwan standard)",
+  "name": "Simplified Chinese to Traditional Chinese (Taiwan Standard)",
+  "normalization": [
+    { "dict": { "type": "ocd2", "file": "CJK_Compatibility_Ideographs.ocd2" } }
+  ],
   "segmentation": {
     "type": "mmseg",
-    "dict": {
-      "type": "ocd2",
-      "file": "STPhrases.ocd2"
-    }
-  },
-  "conversion_chain": [{
     "dict": {
       "type": "group",
-      "dicts": [{
-        "type": "ocd2",
-        "file": "STPhrases.ocd2"
-      }, {
-        "type": "ocd2",
-        "file": "STCharacters.ocd2"
-      }]
+      "match_policy": "union",
+      "dicts": [
+        { "type": "ocd2", "file": "STPhrases.ocd2" },
+        { "type": "ocd2", "file": "STPhrases_GeneratedFromRegionalPhrases.ocd2" }
+      ]
     }
-  }, {
-    "dict": {
-      "type": "ocd2",
-      "file": "TWVariants.ocd2"
+  },
+  "conversion_chain": [
+    {
+      "dict": {
+        "type": "group",
+        "match_policy": "short_circuit",
+        "dicts": [
+          {
+            "type": "group",
+            "match_policy": "union",
+            "dicts": [
+              { "type": "ocd2", "file": "STPhrases.ocd2" },
+              { "type": "ocd2", "file": "STPhrases_GeneratedFromRegionalPhrases.ocd2" }
+            ]
+          },
+          { "type": "ocd2", "file": "STCharacters.ocd2" }
+        ]
+      }
+    },
+    {
+      "dict": {
+        "type": "group",
+        "match_policy": "short_circuit",
+        "dicts": [
+          { "type": "ocd2", "file": "TWVariantsPhrases.ocd2" },
+          { "type": "ocd2", "file": "TWVariants.ocd2" }
+        ]
+      }
     }
-  }]
+  ]
 }