allow inflection rule chains with "polite" (#627)

* fix deinflection bug * add test * fix polite te and past * expand verb flag * fix polite negative * fix polite volitional * fix polite past negative * remove unneeded rules in * new deinflector format * make nicer * lint * remove extra line
themoeway · Feb 4, 2024 · 6b327e0 · 6b327e0
1 parent 1a10af1
commit 6b327e0
Show file tree

Hide file tree

Showing 5 changed files with 154 additions and 189 deletions.
diff --git a/ext/data/language/japanese-transforms.json b/ext/data/language/japanese-transforms.json
@@ -1,6 +1,17 @@
 {
     "language": "ja",
     "conditions": {
+        "v": {
+            "name": "Verb",
+            "partsOfSpeech": ["v1", "v5", "vk", "vs", "vz"],
+            "i18n": [
+                {
+                    "language": "ja",
+                    "name": "動詞"
+                }
+            ],
+            "subConditions": ["v1", "v5", "vk", "vs", "vz"]
+        },
         "v1": {
             "name": "Ichidan verb",
             "partsOfSpeech": ["v1"],
@@ -365,7 +376,8 @@
                 {"suffixIn": "添うて", "suffixOut": "添う", "conditionsIn": ["iru"], "conditionsOut": ["v5"]},
                 {"suffixIn": "副うて", "suffixOut": "副う", "conditionsIn": ["iru"], "conditionsOut": ["v5"]},
                 {"suffixIn": "厭うて", "suffixOut": "厭う", "conditionsIn": ["iru"], "conditionsOut": ["v5"]},
-                {"suffixIn": "のたもうて", "suffixOut": "のたまう", "conditionsIn": ["iru"], "conditionsOut": ["v5"]}
+                {"suffixIn": "のたもうて", "suffixOut": "のたまう", "conditionsIn": ["iru"], "conditionsOut": ["v5"]},
+                {"suffixIn": "まして", "suffixOut": "ます", "conditionsIn": [], "conditionsOut": ["v"]}
             ]
         },
         {
@@ -531,7 +543,7 @@
         {
             "name": "imperative negative",
             "rules": [
-                {"suffixIn": "な", "suffixOut": "", "conditionsIn": [], "conditionsOut": ["v1", "v5", "vk", "vs", "vz"]}
+                {"suffixIn": "な", "suffixOut": "", "conditionsIn": [], "conditionsOut": ["v"]}
             ]
         },
         {
@@ -592,7 +604,8 @@
                 {"suffixIn": "為ない", "suffixOut": "為る", "conditionsIn": ["adj-i"], "conditionsOut": ["vs"]},
                 {"suffixIn": "こない", "suffixOut": "くる", "conditionsIn": ["adj-i"], "conditionsOut": ["vk"]},
                 {"suffixIn": "来ない", "suffixOut": "来る", "conditionsIn": ["adj-i"], "conditionsOut": ["vk"]},
-                {"suffixIn": "來ない", "suffixOut": "來る", "conditionsIn": ["adj-i"], "conditionsOut": ["vk"]}
+                {"suffixIn": "來ない", "suffixOut": "來る", "conditionsIn": ["adj-i"], "conditionsOut": ["vk"]},
+                {"suffixIn": "ません", "suffixOut": "ます", "conditionsIn": ["v"], "conditionsOut": ["v"]}
             ]
         },
         {
@@ -659,114 +672,31 @@
                 {"suffixIn": "添うた", "suffixOut": "添う", "conditionsIn": [], "conditionsOut": ["v5"]},
                 {"suffixIn": "副うた", "suffixOut": "副う", "conditionsIn": [], "conditionsOut": ["v5"]},
                 {"suffixIn": "厭うた", "suffixOut": "厭う", "conditionsIn": [], "conditionsOut": ["v5"]},
-                {"suffixIn": "のたもうた", "suffixOut": "のたまう", "conditionsIn": [], "conditionsOut": ["v5"]}
+                {"suffixIn": "のたもうた", "suffixOut": "のたまう", "conditionsIn": [], "conditionsOut": ["v5"]},
+                {"suffixIn": "ました", "suffixOut": "ます", "conditionsIn": [], "conditionsOut": ["v"]},
+                {"suffixIn": "ませんでした", "suffixOut": "ません", "conditionsIn": [], "conditionsOut": ["v"]}
             ]
         },
         {
             "name": "polite",
             "rules": [
-                {"suffixIn": "ます", "suffixOut": "る", "conditionsIn": [], "conditionsOut": ["v1"]},
-                {"suffixIn": "います", "suffixOut": "う", "conditionsIn": [], "conditionsOut": ["v5"]},
-                {"suffixIn": "きます", "suffixOut": "く", "conditionsIn": [], "conditionsOut": ["v5"]},
-                {"suffixIn": "ぎます", "suffixOut": "ぐ", "conditionsIn": [], "conditionsOut": ["v5"]},
-                {"suffixIn": "します", "suffixOut": "す", "conditionsIn": [], "conditionsOut": ["v5"]},
-                {"suffixIn": "ちます", "suffixOut": "つ", "conditionsIn": [], "conditionsOut": ["v5"]},
-                {"suffixIn": "にます", "suffixOut": "ぬ", "conditionsIn": [], "conditionsOut": ["v5"]},
-                {"suffixIn": "びます", "suffixOut": "ぶ", "conditionsIn": [], "conditionsOut": ["v5"]},
-                {"suffixIn": "みます", "suffixOut": "む", "conditionsIn": [], "conditionsOut": ["v5"]},
-                {"suffixIn": "ります", "suffixOut": "る", "conditionsIn": [], "conditionsOut": ["v5"]},
-                {"suffixIn": "じます", "suffixOut": "ずる", "conditionsIn": [], "conditionsOut": ["vz"]},
-                {"suffixIn": "します", "suffixOut": "する", "conditionsIn": [], "conditionsOut": ["vs"]},
-                {"suffixIn": "為ます", "suffixOut": "為る", "conditionsIn": [], "conditionsOut": ["vs"]},
-                {"suffixIn": "きます", "suffixOut": "くる", "conditionsIn": [], "conditionsOut": ["vk"]},
-                {"suffixIn": "来ます", "suffixOut": "来る", "conditionsIn": [], "conditionsOut": ["vk"]},
-                {"suffixIn": "來ます", "suffixOut": "來る", "conditionsIn": [], "conditionsOut": ["vk"]}
-            ]
-        },
-        {
-            "name": "polite negative",
-            "rules": [
-                {"suffixIn": "くありません", "suffixOut": "い", "conditionsIn": [], "conditionsOut": ["adj-i"]},
-                {"suffixIn": "ません", "suffixOut": "る", "conditionsIn": [], "conditionsOut": ["v1"]},
-                {"suffixIn": "いません", "suffixOut": "う", "conditionsIn": [], "conditionsOut": ["v5"]},
-                {"suffixIn": "きません", "suffixOut": "く", "conditionsIn": [], "conditionsOut": ["v5"]},
-                {"suffixIn": "ぎません", "suffixOut": "ぐ", "conditionsIn": [], "conditionsOut": ["v5"]},
-                {"suffixIn": "しません", "suffixOut": "す", "conditionsIn": [], "conditionsOut": ["v5"]},
-                {"suffixIn": "ちません", "suffixOut": "つ", "conditionsIn": [], "conditionsOut": ["v5"]},
-                {"suffixIn": "にません", "suffixOut": "ぬ", "conditionsIn": [], "conditionsOut": ["v5"]},
-                {"suffixIn": "びません", "suffixOut": "ぶ", "conditionsIn": [], "conditionsOut": ["v5"]},
-                {"suffixIn": "みません", "suffixOut": "む", "conditionsIn": [], "conditionsOut": ["v5"]},
-                {"suffixIn": "りません", "suffixOut": "る", "conditionsIn": [], "conditionsOut": ["v5"]},
-                {"suffixIn": "じません", "suffixOut": "ずる", "conditionsIn": [], "conditionsOut": ["vz"]},
-                {"suffixIn": "しません", "suffixOut": "する", "conditionsIn": [], "conditionsOut": ["vs"]},
-                {"suffixIn": "為ません", "suffixOut": "為る", "conditionsIn": [], "conditionsOut": ["vs"]},
-                {"suffixIn": "きません", "suffixOut": "くる", "conditionsIn": [], "conditionsOut": ["vk"]},
-                {"suffixIn": "来ません", "suffixOut": "来る", "conditionsIn": [], "conditionsOut": ["vk"]},
-                {"suffixIn": "來ません", "suffixOut": "來る", "conditionsIn": [], "conditionsOut": ["vk"]}
-            ]
-        },
-        {
-            "name": "polite past",
-            "rules": [
-                {"suffixIn": "ました", "suffixOut": "る", "conditionsIn": [], "conditionsOut": ["v1"]},
-                {"suffixIn": "いました", "suffixOut": "う", "conditionsIn": [], "conditionsOut": ["v5"]},
-                {"suffixIn": "きました", "suffixOut": "く", "conditionsIn": [], "conditionsOut": ["v5"]},
-                {"suffixIn": "ぎました", "suffixOut": "ぐ", "conditionsIn": [], "conditionsOut": ["v5"]},
-                {"suffixIn": "しました", "suffixOut": "す", "conditionsIn": [], "conditionsOut": ["v5"]},
-                {"suffixIn": "ちました", "suffixOut": "つ", "conditionsIn": [], "conditionsOut": ["v5"]},
-                {"suffixIn": "にました", "suffixOut": "ぬ", "conditionsIn": [], "conditionsOut": ["v5"]},
-                {"suffixIn": "びました", "suffixOut": "ぶ", "conditionsIn": [], "conditionsOut": ["v5"]},
-                {"suffixIn": "みました", "suffixOut": "む", "conditionsIn": [], "conditionsOut": ["v5"]},
-                {"suffixIn": "りました", "suffixOut": "る", "conditionsIn": [], "conditionsOut": ["v5"]},
-                {"suffixIn": "じました", "suffixOut": "ずる", "conditionsIn": [], "conditionsOut": ["vz"]},
-                {"suffixIn": "しました", "suffixOut": "する", "conditionsIn": [], "conditionsOut": ["vs"]},
-                {"suffixIn": "為ました", "suffixOut": "為る", "conditionsIn": [], "conditionsOut": ["vs"]},
-                {"suffixIn": "きました", "suffixOut": "くる", "conditionsIn": [], "conditionsOut": ["vk"]},
-                {"suffixIn": "来ました", "suffixOut": "来る", "conditionsIn": [], "conditionsOut": ["vk"]},
-                {"suffixIn": "來ました", "suffixOut": "來る", "conditionsIn": [], "conditionsOut": ["vk"]}
-            ]
-        },
-        {
-            "name": "polite past negative",
-            "rules": [
-                {"suffixIn": "くありませんでした", "suffixOut": "い", "conditionsIn": [], "conditionsOut": ["adj-i"]},
-                {"suffixIn": "ませんでした", "suffixOut": "る", "conditionsIn": [], "conditionsOut": ["v1"]},
-                {"suffixIn": "いませんでした", "suffixOut": "う", "conditionsIn": [], "conditionsOut": ["v5"]},
-                {"suffixIn": "きませんでした", "suffixOut": "く", "conditionsIn": [], "conditionsOut": ["v5"]},
-                {"suffixIn": "ぎませんでした", "suffixOut": "ぐ", "conditionsIn": [], "conditionsOut": ["v5"]},
-                {"suffixIn": "しませんでした", "suffixOut": "す", "conditionsIn": [], "conditionsOut": ["v5"]},
-                {"suffixIn": "ちませんでした", "suffixOut": "つ", "conditionsIn": [], "conditionsOut": ["v5"]},
-                {"suffixIn": "にませんでした", "suffixOut": "ぬ", "conditionsIn": [], "conditionsOut": ["v5"]},
-                {"suffixIn": "びませんでした", "suffixOut": "ぶ", "conditionsIn": [], "conditionsOut": ["v5"]},
-                {"suffixIn": "みませんでした", "suffixOut": "む", "conditionsIn": [], "conditionsOut": ["v5"]},
-                {"suffixIn": "りませんでした", "suffixOut": "る", "conditionsIn": [], "conditionsOut": ["v5"]},
-                {"suffixIn": "じませんでした", "suffixOut": "ずる", "conditionsIn": [], "conditionsOut": ["vz"]},
-                {"suffixIn": "しませんでした", "suffixOut": "する", "conditionsIn": [], "conditionsOut": ["vs"]},
-                {"suffixIn": "為ませんでした", "suffixOut": "為る", "conditionsIn": [], "conditionsOut": ["vs"]},
-                {"suffixIn": "きませんでした", "suffixOut": "くる", "conditionsIn": [], "conditionsOut": ["vk"]},
-                {"suffixIn": "来ませんでした", "suffixOut": "来る", "conditionsIn": [], "conditionsOut": ["vk"]},
-                {"suffixIn": "來ませんでした", "suffixOut": "來る", "conditionsIn": [], "conditionsOut": ["vk"]}
-            ]
-        },
-        {
-            "name": "polite volitional",
-            "rules": [
-                {"suffixIn": "ましょう", "suffixOut": "る", "conditionsIn": [], "conditionsOut": ["v1"]},
-                {"suffixIn": "いましょう", "suffixOut": "う", "conditionsIn": [], "conditionsOut": ["v5"]},
-                {"suffixIn": "きましょう", "suffixOut": "く", "conditionsIn": [], "conditionsOut": ["v5"]},
-                {"suffixIn": "ぎましょう", "suffixOut": "ぐ", "conditionsIn": [], "conditionsOut": ["v5"]},
-                {"suffixIn": "しましょう", "suffixOut": "す", "conditionsIn": [], "conditionsOut": ["v5"]},
-                {"suffixIn": "ちましょう", "suffixOut": "つ", "conditionsIn": [], "conditionsOut": ["v5"]},
-                {"suffixIn": "にましょう", "suffixOut": "ぬ", "conditionsIn": [], "conditionsOut": ["v5"]},
-                {"suffixIn": "びましょう", "suffixOut": "ぶ", "conditionsIn": [], "conditionsOut": ["v5"]},
-                {"suffixIn": "みましょう", "suffixOut": "む", "conditionsIn": [], "conditionsOut": ["v5"]},
-                {"suffixIn": "りましょう", "suffixOut": "る", "conditionsIn": [], "conditionsOut": ["v5"]},
-                {"suffixIn": "じましょう", "suffixOut": "ずる", "conditionsIn": [], "conditionsOut": ["vz"]},
-                {"suffixIn": "しましょう", "suffixOut": "する", "conditionsIn": [], "conditionsOut": ["vs"]},
-                {"suffixIn": "為ましょう", "suffixOut": "為る", "conditionsIn": [], "conditionsOut": ["vs"]},
-                {"suffixIn": "きましょう", "suffixOut": "くる", "conditionsIn": [], "conditionsOut": ["vk"]},
-                {"suffixIn": "来ましょう", "suffixOut": "来る", "conditionsIn": [], "conditionsOut": ["vk"]},
-                {"suffixIn": "來ましょう", "suffixOut": "來る", "conditionsIn": [], "conditionsOut": ["vk"]}
+                {"suffixIn": "ます", "suffixOut": "る", "conditionsIn": ["v1"], "conditionsOut": ["v1"]},
+                {"suffixIn": "います", "suffixOut": "う", "conditionsIn": ["v5"], "conditionsOut": ["v5"]},
+                {"suffixIn": "きます", "suffixOut": "く", "conditionsIn": ["v5"], "conditionsOut": ["v5"]},
+                {"suffixIn": "ぎます", "suffixOut": "ぐ", "conditionsIn": ["v5"], "conditionsOut": ["v5"]},
+                {"suffixIn": "します", "suffixOut": "す", "conditionsIn": ["v5"], "conditionsOut": ["v5"]},
+                {"suffixIn": "ちます", "suffixOut": "つ", "conditionsIn": ["v5"], "conditionsOut": ["v5"]},
+                {"suffixIn": "にます", "suffixOut": "ぬ", "conditionsIn": ["v5"], "conditionsOut": ["v5"]},
+                {"suffixIn": "びます", "suffixOut": "ぶ", "conditionsIn": ["v5"], "conditionsOut": ["v5"]},
+                {"suffixIn": "みます", "suffixOut": "む", "conditionsIn": ["v5"], "conditionsOut": ["v5"]},
+                {"suffixIn": "ります", "suffixOut": "る", "conditionsIn": ["v5"], "conditionsOut": ["v5"]},
+                {"suffixIn": "じます", "suffixOut": "ずる", "conditionsIn": ["vz"], "conditionsOut": ["vz"]},
+                {"suffixIn": "します", "suffixOut": "する", "conditionsIn": ["vs"], "conditionsOut": ["vs"]},
+                {"suffixIn": "為ます", "suffixOut": "為る", "conditionsIn": ["vs"], "conditionsOut": ["vs"]},
+                {"suffixIn": "きます", "suffixOut": "くる", "conditionsIn": ["vk"], "conditionsOut": ["vk"]},
+                {"suffixIn": "来ます", "suffixOut": "来る", "conditionsIn": ["vk"], "conditionsOut": ["vk"]},
+                {"suffixIn": "來ます", "suffixOut": "來る", "conditionsIn": ["vk"], "conditionsOut": ["vk"]},
+                {"suffixIn": "くあります", "suffixOut": "い", "conditionsIn": ["v"], "conditionsOut": ["adj-i"]}
             ]
         },
         {
@@ -819,7 +749,8 @@
                 {"suffixIn": "為よう", "suffixOut": "為る", "conditionsIn": [], "conditionsOut": ["vs"]},
                 {"suffixIn": "こよう", "suffixOut": "くる", "conditionsIn": [], "conditionsOut": ["vk"]},
                 {"suffixIn": "来よう", "suffixOut": "来る", "conditionsIn": [], "conditionsOut": ["vk"]},
-                {"suffixIn": "來よう", "suffixOut": "來る", "conditionsIn": [], "conditionsOut": ["vk"]}
+                {"suffixIn": "來よう", "suffixOut": "來る", "conditionsIn": [], "conditionsOut": ["vk"]},
+                {"suffixIn": "ましょう", "suffixOut": "ます", "conditionsIn": [], "conditionsOut": ["v"]}
             ]
         },
         {

diff --git a/test/data/anki-note-builder-test-results.json b/test/data/anki-note-builder-test-results.json
@@ -1692,7 +1692,7 @@
         "cloze-body": "打ち込んでいませんでした",
         "cloze-prefix": "cloze-prefix",
         "cloze-suffix": "cloze-suffix",
-        "conjugation": "-te « progressive or perfect « polite past negative",
+        "conjugation": "-te « progressive or perfect « polite « negative « past",
         "dictionary": "Test Dictionary 2",
         "document-title": "title",
         "expression": "打ち込む",
@@ -1727,7 +1727,7 @@
         "cloze-body": "打ち込んでいませんでした",
         "cloze-prefix": "cloze-prefix",
         "cloze-suffix": "cloze-suffix",
-        "conjugation": "-te « progressive or perfect « polite past negative",
+        "conjugation": "-te « progressive or perfect « polite « negative « past",
         "dictionary": "Test Dictionary 2",
         "document-title": "title",
         "expression": "打ち込む",
@@ -1762,7 +1762,7 @@
         "cloze-body": "打ち込んでいませんでした",
         "cloze-prefix": "cloze-prefix",
         "cloze-suffix": "cloze-suffix",
-        "conjugation": "-te « progressive or perfect « polite past negative",
+        "conjugation": "-te « progressive or perfect « polite « negative « past",
         "dictionary": "Test Dictionary 2",
         "document-title": "title",
         "expression": "打ち込む",
@@ -1797,7 +1797,7 @@
         "cloze-body": "打ち込んでいませんでした",
         "cloze-prefix": "cloze-prefix",
         "cloze-suffix": "cloze-suffix",
-        "conjugation": "-te « progressive or perfect « polite past negative",
+        "conjugation": "-te « progressive or perfect « polite « negative « past",
         "dictionary": "Test Dictionary 2",
         "document-title": "title",
         "expression": "打ち込む",
@@ -2837,7 +2837,7 @@
         "cloze-body": "testました",
         "cloze-prefix": "cloze-prefix",
         "cloze-suffix": "cloze-suffix",
-        "conjugation": "polite past",
+        "conjugation": "polite « past",
         "dictionary": "Test Dictionary 2",
         "document-title": "title",
         "expression": "読む",

diff --git a/test/data/translator-test-results-note-data1.json b/test/data/translator-test-results-note-data1.json
@@ -15851,7 +15851,9 @@
               "inflectionRules": [
                 "-te",
                 "progressive or perfect",
-                "polite past negative"
+                "polite",
+                "negative",
+                "past"
               ]
             }
           ],
@@ -16276,7 +16278,9 @@
               "inflectionRules": [
                 "-te",
                 "progressive or perfect",
-                "polite past negative"
+                "polite",
+                "negative",
+                "past"
               ]
             }
           ],
@@ -16701,7 +16705,9 @@
               "inflectionRules": [
                 "-te",
                 "progressive or perfect",
-                "polite past negative"
+                "polite",
+                "negative",
+                "past"
               ]
             }
           ],
@@ -17126,7 +17132,9 @@
               "inflectionRules": [
                 "-te",
                 "progressive or perfect",
-                "polite past negative"
+                "polite",
+                "negative",
+                "past"
               ]
             }
           ],
@@ -26937,7 +26945,8 @@
             {
               "source": "algorithm",
               "inflectionRules": [
-                "polite past"
+                "polite",
+                "past"
               ]
             }
           ],