Add "Pitch Accent (Numeric)" field for mining (partially implements #72)

rampaa · Feb 3, 2024 · 4749dce · 4749dce
1 parent 41f48f0
commit 4749dce
Show file tree

Hide file tree

Showing 4 changed files with 95 additions and 0 deletions.
diff --git a/JL.Core/Mining/JLField.cs b/JL.Core/Mining/JLField.cs
@@ -19,6 +19,7 @@ public enum JLField
     [Description("Local Time")] LocalTime,
     Frequencies,
     [Description("Raw Frequencies")] RawFrequencies,
+    [Description("Pitch Accents (Numeric)")] NumericPitchAccents,
 
     // JMdict, JMnedict, KANJIDIC2
     [Description("EDICT ID")] EdictId,

diff --git a/JL.Core/Mining/JLFieldUtils.cs b/JL.Core/Mining/JLFieldUtils.cs
@@ -22,6 +22,7 @@ public static class JLFieldUtils
         JLField.DeconjugationProcess,
         JLField.Frequencies,
         JLField.RawFrequencies,
+        JLField.NumericPitchAccents,
         JLField.EdictId,
         JLField.LocalTime
     };
@@ -46,6 +47,7 @@ public static class JLFieldUtils
         JLField.Sentence,
         JLField.Frequencies,
         JLField.RawFrequencies,
+        JLField.NumericPitchAccents,
         JLField.EdictId,
         JLField.LocalTime
     };
@@ -62,6 +64,7 @@ public static class JLFieldUtils
         JLField.Image,
         JLField.SourceText,
         JLField.Sentence,
+        JLField.NumericPitchAccents,
         JLField.EdictId,
         JLField.LocalTime
     };

diff --git a/JL.Core/Mining/MiningUtils.cs b/JL.Core/Mining/MiningUtils.cs
@@ -2,6 +2,7 @@
 using System.Text;
 using JL.Core.Audio;
 using JL.Core.Dicts;
+using JL.Core.Dicts.PitchAccent;
 using JL.Core.Lookup;
 using JL.Core.Mining.Anki;
 using JL.Core.Network;
@@ -131,9 +132,96 @@ private static Dictionary<JLField, string> GetMiningParameters(LookupResult look
             miningParams[JLField.RadicalNames] = string.Join(", ", lookupResult.RadicalNames);
         }
 
+        if (DictUtils.SingleDictTypeDicts.TryGetValue(DictType.PitchAccentYomichan, out Dict? pitchDict))
+        {
+            if (pitchDict.Active)
+            {
+                List<KeyValuePair<string, int>>? pitchAccents = GetPitchAccents(lookupResult.PitchAccentDict ?? pitchDict.Contents, lookupResult);
+                if (pitchAccents is not null)
+                {
+                    StringBuilder numericPitchAccentBuilder = new();
+                    for (int i = 0; i < pitchAccents.Count; i++)
+                    {
+                        KeyValuePair<string, int> pitchAccent = pitchAccents[i];
+                        _ = numericPitchAccentBuilder.Append(CultureInfo.InvariantCulture, $"{pitchAccent.Key}: {pitchAccent.Value}, ");
+                    }
+
+                    _ = numericPitchAccentBuilder.Remove(numericPitchAccentBuilder.Length - 2, 2);
+                    miningParams[JLField.NumericPitchAccents] = numericPitchAccentBuilder.ToString();
+                }
+            }
+        }
+
         return miningParams;
     }
 
+    private static List<KeyValuePair<string, int>>? GetPitchAccents(Dictionary<string, IList<IDictRecord>> pitchDict, LookupResult lookupResult)
+    {
+        List<KeyValuePair<string, int>> pitchAccents = new();
+
+        if (lookupResult.Readings is not null)
+        {
+            for (int i = 0; i < lookupResult.Readings.Length; i++)
+            {
+                string reading = lookupResult.Readings[i];
+                string readingInHiragana = JapaneseUtils.KatakanaToHiragana(reading);
+
+                if (pitchDict.TryGetValue(readingInHiragana, out IList<IDictRecord>? pitchResult))
+                {
+                    foreach (IDictRecord dictRecord in pitchResult)
+                    {
+                        PitchAccentRecord pitchAccentRecord = (PitchAccentRecord)dictRecord;
+                        if (lookupResult.PrimarySpelling == pitchAccentRecord.Spelling
+                            || (lookupResult.AlternativeSpellings?.Contains(pitchAccentRecord.Spelling) ?? false))
+                        {
+                            pitchAccents.Add(KeyValuePair.Create(reading, pitchAccentRecord.Position));
+                            break;
+                        }
+                    }
+                }
+            }
+        }
+
+        else
+        {
+            string primarySpellingInHiragana = JapaneseUtils.KatakanaToHiragana(lookupResult.PrimarySpelling);
+            if (pitchDict.TryGetValue(primarySpellingInHiragana, out IList<IDictRecord>? pitchResult))
+            {
+                foreach (IDictRecord dictRecord in pitchResult)
+                {
+                    PitchAccentRecord pitchAccentRecord = (PitchAccentRecord)dictRecord;
+                    if (pitchAccentRecord.Reading is null)
+                    {
+                        pitchAccents.Add(KeyValuePair.Create(lookupResult.PrimarySpelling, pitchAccentRecord.Position));
+                        break;
+                    }
+                }
+            }
+
+            else if (lookupResult.AlternativeSpellings is not null)
+            {
+                for (int i = 0; i < lookupResult.AlternativeSpellings.Length; i++)
+                {
+                    string alternativeSpellingInHiragana = JapaneseUtils.KatakanaToHiragana(lookupResult.AlternativeSpellings[i]);
+                    if (pitchDict.TryGetValue(alternativeSpellingInHiragana, out pitchResult))
+                    {
+                        foreach (IDictRecord dictRecord in pitchResult)
+                        {
+                            PitchAccentRecord pitchAccentRecord = (PitchAccentRecord)dictRecord;
+                            if (pitchAccentRecord.Reading is null)
+                            {
+                                pitchAccents.Add(KeyValuePair.Create(lookupResult.PrimarySpelling, pitchAccentRecord.Position));
+                                break;
+                            }
+                        }
+                    }
+                }
+            }
+        }
+
+        return pitchAccents.Count > 0 ? pitchAccents : null;
+    }
+
     public static async Task MineToFile(LookupResult lookupResult, string currentText, string? selectedDefinitions, int currentCharPosition)
     {
         string filePath;

diff --git a/JL.Windows/GUI/PreferencesWindow.xaml.cs b/JL.Windows/GUI/PreferencesWindow.xaml.cs
@@ -63,6 +63,7 @@ public static bool IsItVisible()
                                             • Deconjugation Process: Deconjugation path from the "Matched Text" to "Deconjugated Matched Text".
                                             • Frequencies: Frequency info for the mined word, e.g., "VN: 77, jpdb: 666".
                                             • Raw Frequencies: Raw frequency info for the mined word, e.g., "77, 666".
+                                            • Pitch Accents (Numeric): Pitch accents for the mined word in numeric form, e.g., おんな: 3, おみな: 0, おうな: 1
                                             • EDICT ID: JMDict entry ID.
                                             • Local Time: Mining date and time expressed in local timezone.
                                             """;
@@ -84,6 +85,7 @@ public static bool IsItVisible()
                                              • Sentence: Sentence in which the mined kanji appears in.
                                              • Frequencies: Frequency info for the kanji, e.g., "KANJIDIC2: 77, jpdb: 666".
                                              • Raw Frequencies: Raw frequency info for the mined word, e.g., "77, 666".
+                                             • Pitch Accents (Numeric): Pitch accents for the mined word in numeric form, e.g., おんな: 3, おみな: 0, おうな: 1
                                              • EDICT ID: KANJIDIC2 entry ID.
                                              • Local Time: Mining date and time expressed in local timezone.
                                              """;
@@ -97,6 +99,7 @@ public static bool IsItVisible()
                                             • Dictionary Name: Name of the dictionary, e.g., "JMnedict".
                                             • Source Text: Whole text in which the mined name appears in.
                                             • Sentence: Sentence in which the mined name appears in.
+                                            • Pitch Accents (Numeric): Pitch accents for the mined word in numeric form, e.g., おんな: 3, おみな: 0, おうな: 1
                                             • EDICT ID: JMnedict entry ID.
                                             • Local Time: Mining date and time expressed in local timezone.
                                             """;