fix missing IPA tags (#168)

yomidevs · Nov 14, 2024 · 351fa0d · 351fa0d
1 parent ca096e5
commit 351fa0d
Show file tree

Hide file tree

Showing 14 changed files with 321 additions and 18 deletions.
diff --git a/3-tidy-up.js b/3-tidy-up.js
@@ -278,8 +278,11 @@ function processForms(forms, word, pos) {
 function saveIpaResult(word, readings, pos, ipaObj) {
     for (const reading of readings) {
         const result = lemmaDict[word][reading][pos];
-        if (!result.ipa.some(obj => obj.ipa === ipaObj.ipa)) {
+        const existingIpa = result.ipa.find(obj => obj.ipa === ipaObj.ipa);
+        if (!existingIpa) {
             result.ipa.push(ipaObj);
+        } else {
+            existingIpa.tags = [...new Set([...existingIpa.tags, ...ipaObj.tags])];
         }
     }
 }

diff --git a/data/test/dict/en/en/tag_bank_1.json b/data/test/dict/en/en/tag_bank_1.json
@@ -47,5 +47,12 @@
     -1,
     "noun",
     1
+  ],
+  [
+    "hist",
+    "",
+    0,
+    "historical",
+    0
   ]
 ]
diff --git a/data/test/dict/en/en/term_bank_1.json b/data/test/dict/en/en/term_bank_1.json
@@ -303,5 +303,60 @@
     ],
     0,
     ""
+  ],
+  [
+    "falcon",
+    "",
+    "n",
+    "n",
+    0,
+    [
+      {
+        "type": "structured-content",
+        "content": [
+          {
+            "tag": "div",
+            "content": [
+              "Any bird of the genus Falco, all of which are birds of prey."
+            ]
+          }
+        ]
+      },
+      {
+        "type": "structured-content",
+        "content": [
+          {
+            "tag": "div",
+            "content": [
+              "(falconry) A female such bird, a male being a tiercel."
+            ]
+          }
+        ]
+      }
+    ],
+    0,
+    ""
+  ],
+  [
+    "falcon",
+    "",
+    "hist n",
+    "n",
+    0,
+    [
+      {
+        "type": "structured-content",
+        "content": [
+          {
+            "tag": "div",
+            "content": [
+              "A light cannon used from the 15th to the 17th century; a falconet."
+            ]
+          }
+        ]
+      }
+    ],
+    0,
+    ""
   ]
 ]
diff --git a/data/test/dict/en/en/term_bank_2.json b/data/test/dict/en/en/term_bank_2.json
@@ -96,5 +96,22 @@
     ],
     0,
     ""
+  ],
+  [
+    "falcons",
+    "",
+    "non-lemma",
+    "",
+    0,
+    [
+      [
+        "falcon",
+        [
+          "plural"
+        ]
+      ]
+    ],
+    0,
+    ""
   ]
 ]
diff --git a/data/test/ipa/en/en/tag_bank_1.json b/data/test/ipa/en/en/tag_bank_1.json
@@ -1 +1,30 @@
-[]
+[
+  [
+    "🏴󠁧󠁢󠁥󠁮󠁧󠁿",
+    "dialect",
+    0,
+    "Received-Pronunciation",
+    0
+  ],
+  [
+    "🇺🇸",
+    "dialect",
+    0,
+    "US",
+    0
+  ],
+  [
+    "🇦🇺",
+    "dialect",
+    0,
+    "General-Australian",
+    0
+  ],
+  [
+    "🇳🇿",
+    "dialect",
+    0,
+    "New-Zealand",
+    0
+  ]
+]
diff --git a/data/test/ipa/en/en/term_meta_bank_1.json b/data/test/ipa/en/en/term_meta_bank_1.json
@@ -24,5 +24,52 @@
         }
       ]
     }
+  ],
+  [
+    "falcon",
+    "ipa",
+    {
+      "reading": "falcon",
+      "transcriptions": [
+        {
+          "ipa": "/ˈfɔː(l)kən/",
+          "tags": [
+            "🏴󠁧󠁢󠁥󠁮󠁧󠁿"
+          ]
+        },
+        {
+          "ipa": "/ˈfɒlkən/",
+          "tags": [
+            "🏴󠁧󠁢󠁥󠁮󠁧󠁿"
+          ]
+        },
+        {
+          "ipa": "/ˈfælkən/",
+          "tags": [
+            "🇺🇸",
+            "🇦🇺"
+          ]
+        },
+        {
+          "ipa": "/ˈfɑːlkən/",
+          "tags": [
+            "🇺🇸"
+          ]
+        },
+        {
+          "ipa": "/ˈfoːlkən/",
+          "tags": [
+            "🇦🇺",
+            "🇳🇿"
+          ]
+        },
+        {
+          "ipa": "/ˈfɛlkən/",
+          "tags": [
+            "🇳🇿"
+          ]
+        }
+      ]
+    }
   ]
 ]
diff --git a/data/test/ipa/fa/en/tag_bank_1.json b/data/test/ipa/fa/en/tag_bank_1.json
@@ -7,10 +7,10 @@
     0
   ],
   [
-    "🇦🇫",
+    "🇹🇯",
     "dialect",
     0,
-    "Dari",
+    "Tajik",
     0
   ],
   [
@@ -21,17 +21,24 @@
     0
   ],
   [
-    "🇮🇷",
+    "🇦🇫",
     "dialect",
     0,
-    "Iran",
+    "Dari",
     0
   ],
   [
-    "🇹🇯",
+    "🇦🇫KA",
     "dialect",
     0,
-    "Tajik",
+    "Kabuli",
+    0
+  ],
+  [
+    "🇮🇷",
+    "dialect",
+    0,
+    "Iran",
     0
   ]
 ]
diff --git a/data/test/ipa/fa/en/term_meta_bank_1.json b/data/test/ipa/fa/en/term_meta_bank_1.json
@@ -8,14 +8,18 @@
         {
           "ipa": "[mulk]",
           "tags": [
-            "classic"
+            "classic",
+            "Hazaragi",
+            "🇹🇯",
+            "frm"
           ]
         },
         {
           "ipa": "[mʊlk]",
           "tags": [
             "🇦🇫",
-            "frm"
+            "frm",
+            "🇦🇫KA"
           ]
         },
         {
@@ -44,7 +48,8 @@
           "ipa": "[fɑːɾ.síː]",
           "tags": [
             "🇦🇫",
-            "frm"
+            "frm",
+            "🇦🇫KA"
           ]
         },
         {

diff --git a/data/test/ipa/la/en/term_meta_bank_1.json b/data/test/ipa/la/en/term_meta_bank_1.json
@@ -14,7 +14,8 @@
         {
           "ipa": "[ˈfäːmä]",
           "tags": [
-            "Classical"
+            "Classical",
+            "✝️"
           ]
         },
         {
@@ -179,7 +180,8 @@
         {
           "ipa": "/ˈdo.mus/",
           "tags": [
-            "Classical"
+            "Classical",
+            "✝️"
           ]
         },
         {

diff --git a/data/test/kaikki/en-en.json b/data/test/kaikki/en-en.json
diff --git a/data/test/tidy/en-en-forms-0.json b/data/test/tidy/en-en-forms-0.json
@@ -87,6 +87,28 @@
           ]
         ]
       }
+    ],
+    [
+      "falcon",
+      {
+        "_type": "map",
+        "map": [
+          [
+            "falcons",
+            {
+              "_type": "map",
+              "map": [
+                [
+                  "noun",
+                  [
+                    "plural"
+                  ]
+                ]
+              ]
+            }
+          ]
+        ]
+      }
     ]
   ]
 }
-Original file line number
+Diff line change
@@ Expand Up / @@ -47,5 +47,12 @@ @@
         -1,
         "noun",
+      ],
+      [
+        "hist",
+        "",
+,
+        "historical",
       ]
     ]