use regular space in multiword inflections, [de-de] parse more inflec…

…tion glosses (#126) * [de] add past participle to multiword inflections * [de] add present participle to multiword inflections * add demonstrativpronomens deinflections * fix multiword * add pronomens
yomidevs · Aug 7, 2024 · 53a8943 · 53a8943
1 parent 03940a0
commit 53a8943
Show file tree

Hide file tree

Showing 5 changed files with 33 additions and 26 deletions.
diff --git a/3-tidy-up.js b/3-tidy-up.js
@@ -23,7 +23,7 @@ function isInflectionGloss(glosses, formOf) {
     glossesString = JSON.stringify(glosses);
     switch (targetIso) {
         case 'de':
-            if (glosses.some(gloss => /des (?:Verbs|Adjektivs|Substantivs)/.test(gloss))) return true;
+            if (glosses.some(gloss => /des (?:Verbs|Adjektivs|Substantivs|Demonstrativpronomens|Possessivpronomens|Pronomens)/.test(gloss))) return true;
         case 'en':
             if (glosses.some(gloss => /.*inflection of.*/.test(gloss))) return true;
             if(!Array.isArray(formOf)) return false;
@@ -296,7 +296,7 @@ function processInflectionGlosses(glosses, word, pos) {
 }
 
 function processGermanInflectionGlosses(glosses, word, pos) {
-    const match1 = glosses[0].match(/(.*)des (?:Verbs|Adjektivs|Substantivs) (.*)$/);
+    const match1 = glosses[0].match(/(.*)des (?:Verbs|Adjektivs|Substantivs|Demonstrativpronomens|Possessivpronomens|Pronomens) (.*)$/);
     if (!match1 || match1.length < 3) return;
     const inflection = match1[1].trim();
     const lemma = match1[2].trim();

diff --git a/4-make-yomitan.js b/4-make-yomitan.js
@@ -316,9 +316,16 @@ let lastTermBankIndex = 0;
                         hypotheses = hypotheses
                             .map((hypothesis) => 
                                 hypothesis
-                                    .map((inflection) => (inflection).trim())
+                                    .map((inflection) => inflection.trim())
                                     .filter(Boolean)
-                            ).filter(hypothesis => hypothesis.length);
+                            )
+                            .filter(hypothesis => hypothesis.length)
+                            .map((hypothesis) => 
+                                hypothesis.map((inflection) => 
+                                    inflection.replace(/\u00A0/g, ' ')
+                                )
+                            );
+
 
                         return hypotheses;
                     });

diff --git a/data/test/dict/cs/en/term_bank_2.json b/data/test/dict/cs/en/term_bank_2.json
@@ -11,15 +11,15 @@
         [
           "singular",
           "feminine",
-          "past active participle"
+          "past active participle"
         ]
       ],
       [
         "přít",
         [
           "plural",
           "neuter",
-          "past active participle"
+          "past active participle"
         ]
       ]
     ],
@@ -38,7 +38,7 @@
         [
           "first-person",
           "singular",
-          "present indicative"
+          "present indicative"
         ]
       ]
     ],
@@ -66,7 +66,7 @@
         [
           "first-person",
           "plural",
-          "present indicative"
+          "present indicative"
         ]
       ]
     ],
@@ -85,7 +85,7 @@
         [
           "second-person",
           "singular",
-          "present indicative"
+          "present indicative"
         ]
       ]
     ],
@@ -113,7 +113,7 @@
         [
           "second-person",
           "plural",
-          "present indicative"
+          "present indicative"
         ]
       ]
     ],
@@ -152,15 +152,15 @@
         [
           "singular",
           "masculine",
-          "present transgressive"
+          "present transgressive"
         ]
       ],
       [
         "přít",
         [
           "third-person",
           "singular",
-          "present indicative"
+          "present indicative"
         ]
       ]
     ],
@@ -179,7 +179,7 @@
         [
           "third-person",
           "plural",
-          "present indicative"
+          "present indicative"
         ]
       ]
     ],
@@ -199,7 +199,7 @@
           "singular",
           "past",
           "participle",
-          "masculine animate"
+          "masculine animate"
         ]
       ],
       [
@@ -208,7 +208,7 @@
           "singular",
           "past",
           "participle",
-          "masculine inanimate"
+          "masculine inanimate"
         ]
       ]
     ],
@@ -228,7 +228,7 @@
           "plural",
           "past",
           "participle",
-          "masculine animate"
+          "masculine animate"
         ]
       ]
     ],
@@ -257,7 +257,7 @@
           "plural",
           "past",
           "participle",
-          "masculine inanimate"
+          "masculine inanimate"
         ]
       ]
     ],
@@ -296,7 +296,7 @@
         [
           "singular",
           "masculine",
-          "present transgressive"
+          "present transgressive"
         ]
       ]
     ],
@@ -316,7 +316,7 @@
           "singular",
           "feminine",
           "neuter",
-          "present transgressive"
+          "present transgressive"
         ]
       ]
     ],
@@ -336,7 +336,7 @@
           "singular",
           "feminine",
           "neuter",
-          "present transgressive"
+          "present transgressive"
         ]
       ]
     ],
@@ -354,7 +354,7 @@
         "přít",
         [
           "plural",
-          "present transgressive"
+          "present transgressive"
         ]
       ]
     ],
@@ -372,7 +372,7 @@
         "přít",
         [
           "plural",
-          "present transgressive"
+          "present transgressive"
         ]
       ]
     ],

diff --git a/data/test/dict/de/en/term_bank_2.json b/data/test/dict/de/en/term_bank_2.json
@@ -89,7 +89,7 @@
       [
         "pflegen",
         [
-          "past participle"
+          "past participle"
         ]
       ]
     ],
@@ -149,7 +149,7 @@
       [
         "pflegen",
         [
-          "past participle"
+          "past participle"
         ]
       ]
     ],
@@ -193,7 +193,7 @@
       [
         "pflegen",
         [
-          "present participle"
+          "present participle"
         ]
       ]
     ],

diff --git a/data/test/dict/la/en/term_bank_2.json b/data/test/dict/la/en/term_bank_2.json
@@ -2551,7 +2551,7 @@
       [
         "legō",
         [
-          "perfect passive participle"
+          "perfect passive participle"
         ]
       ]
     ],
-Original file line number
+Diff line change
@@ Expand Up / @@ -2551,7 +2551,7 @@ @@
           [
             "legō",
             [
-              "perfect passive participle"
+              "perfect passive participle"
             ]
           ]
         ],
@@ Expand Down @@