fix multiple slavic language issues, minor cosmetic change (#180)

* improve russian form handling, write tests Fixed Russian forms being improperly marked as lemmas. Wrote tests. * add missing common russian case, write tests * filter `dated` inflection tag, add test * improve noun gender coloring Changed masculine, feminine, and neuter tag coloring to something more visually pleasing. * add missing noun gender and verb aspect Extracts noun gender and verb aspect from headword `expansion`. Should work excellent for Slavic languages, and even Latin.
yomidevs · Dec 20, 2024 · 0814121 · 0814121
1 parent 6e1850f
commit 0814121
Show file tree

Hide file tree

Showing 15 changed files with 2,314 additions and 21 deletions.
diff --git a/3-tidy-up.js b/3-tidy-up.js
@@ -43,7 +43,7 @@ function isInflectionGloss(glosses, formOf) {
             if(!Array.isArray(formOf)) return false;
             for (const {word: lemma} of formOf) {
                 if(!lemma) continue;
-                if (glosses.some(gloss => new RegExp(`of ${escapeRegExp(lemma)}$`).test(gloss))) return true;
+                if (glosses.some(gloss => new RegExp(`of ${escapeRegExp(lemma)}($| \(.+?\)$)`).test(gloss))) return true;
             }
 
         case 'fr':
@@ -93,7 +93,8 @@ const blacklistedTags = [
     'obsolete',
     'archaic',
     'used-in-the-form',
-    'romanization'
+    'romanization',
+    'dated'
 ];
 
 const identityTags = [
@@ -131,7 +132,7 @@ function handleLine(parsedLine) {
 
     processForms(forms, word, pos);
 
-    const {senses} = parsedLine;
+    const {senses, head_templates} = parsedLine;
     if (!senses) return;
 
     /** @type {IpaInfo[]} */
@@ -164,6 +165,31 @@ function handleLine(parsedLine) {
             tags.push(...sense.raw_tags);
         }
 
+        if (head_templates && targetIso === 'en') {
+            const tagMatch = [
+                ['pf', 'perfective'],
+                ['impf', 'imperfective'],
+                ['m', 'masculine'],
+                ['f', 'feminine'],
+                ['n', 'neuter'],
+                ['inan', 'inanimate'],
+                ['anim', 'animate'],
+            ];
+
+            for (const entry of head_templates) {
+                if (entry.expansion) {
+                    for (const [match, tag] of tagMatch) {
+                        if (
+                            entry.expansion.replace(/\(.+?\)/g, '').split(' ').includes(match) &&
+                            !tags.includes(tag)
+                        ) {
+                            tags.push(tag);
+                        }
+                    }
+                }
+            }
+        }
+
         return {...sense, glossesArray, tags};
     }));
 

diff --git a/data/language/tag_order.json b/data/language/tag_order.json
@@ -10,7 +10,8 @@
         "accusative",
         "vocative",
         "locative",
-        "instrumental"
+        "instrumental",
+        "prepositional"
     ],
     "persons": [
         "first-person",

diff --git a/data/language/target-language-tags/en/tag_styles.json b/data/language/target-language-tags/en/tag_styles.json
@@ -1,5 +1,5 @@
 {
-    "masc": ".tag[data-details='masculine'] .tag-label{background-color: blue;}",
-    "fem": ".tag[data-details='feminine'] .tag-label{background-color: red;}",
-    "neut": ".tag[data-details='neuter'] .tag-label{background-color: green;}"
+    "masc": ".tag[data-details='masculine'] .tag-label{background-color: #4d82e8;}",
+    "fem": ".tag[data-details='feminine'] .tag-label{background-color: #ca4d93;}",
+    "neut": ".tag[data-details='neuter'] .tag-label{background-color: #40ac65;}"
 }
diff --git a/data/test/dict/la/en/tag_bank_1.json b/data/test/dict/la/en/tag_bank_1.json
@@ -6,6 +6,13 @@
     "declension-1",
     0
   ],
+  [
+    "fem",
+    "",
+    -1,
+    "feminine",
+    1
+  ],
   [
     "n",
     "partOfSpeech",
@@ -48,6 +55,13 @@
     "declension-2",
     0
   ],
+  [
+    "neut",
+    "",
+    -1,
+    "neuter",
+    1
+  ],
   [
     "not-comp",
     "",
@@ -76,13 +90,6 @@
     "declension-4",
     0
   ],
-  [
-    "fem",
-    "",
-    -1,
-    "feminine",
-    1
-  ],
   [
     "irreg",
     "",

diff --git a/data/test/dict/la/en/term_bank_1.json b/data/test/dict/la/en/term_bank_1.json
@@ -2,7 +2,7 @@
   [
     "fama",
     "fāma",
-    "decl-1 n",
+    "decl-1 fem n",
     "n",
     0,
     [
@@ -324,7 +324,7 @@
   [
     "lilium",
     "līlium",
-    "decl-2 n",
+    "decl-2 neut n",
     "n",
     0,
     [

diff --git a/data/test/dict/ru/en/tag_bank_1.json b/data/test/dict/ru/en/tag_bank_1.json
@@ -0,0 +1,79 @@
+[
+  [
+    "masc",
+    "",
+    -1,
+    "masculine",
+    1
+  ],
+  [
+    "inanim",
+    "",
+    0,
+    "inanimate",
+    0
+  ],
+  [
+    "n",
+    "partOfSpeech",
+    -1,
+    "noun",
+    1
+  ],
+  [
+    "fig",
+    "",
+    0,
+    "figuratively",
+    0
+  ],
+  [
+    "sl",
+    "",
+    0,
+    "slang",
+    0
+  ],
+  [
+    "pf",
+    "",
+    0,
+    "perfective",
+    0
+  ],
+  [
+    "v",
+    "partOfSpeech",
+    -1,
+    "verb",
+    1
+  ],
+  [
+    "col",
+    "",
+    0,
+    "colloquial",
+    0
+  ],
+  [
+    "impers",
+    "",
+    0,
+    "impersonal",
+    0
+  ],
+  [
+    "reltnl",
+    "",
+    0,
+    "relational",
+    0
+  ],
+  [
+    "adj",
+    "partOfSpeech",
+    -1,
+    "adjective",
+    1
+  ]
+]