feat: operator corrections injected into similar questions

src/main/java/dev/marcospaulo/canalhandia/Corrections.java

@@ -0,0 +1,89 @@
+package dev.marcospaulo.canalhandia;
+
+import org.bukkit.configuration.file.YamlConfiguration;
+
+import java.io.File;
+import java.util.ArrayList;
+import java.util.HashSet;
+import java.util.List;
+import java.util.Locale;
+import java.util.Set;
+
+/**
+ * Operator corrections, injected when a new question resembles one that was
+ * answered wrongly before.
+ *
+ * <p>This is the cheap alternative to fine-tuning: a wrong answer becomes
+ * context, so the same mistake stops recurring.
+ */
+final class Corrections {
+
+    record Entry(String question, String answer) {
+    }
+
+    private final File file;
+    private final List<Entry> entries = new ArrayList<>();
+
+    Corrections(File file) {
+        this.file = file;
+        load();
+    }
+
+    void load() {
+        entries.clear();
+        if (!file.exists()) {
+            return;
+        }
+        YamlConfiguration yaml = YamlConfiguration.loadConfiguration(file);
+        for (String key : yaml.getKeys(false)) {
+            String question = yaml.getString(key + ".pergunta");
+            String answer = yaml.getString(key + ".resposta");
+            if (question != null && answer != null) {
+                entries.add(new Entry(question, answer));
+            }
+        }
+    }
+
+    void add(String question, String answer) {
+        entries.add(new Entry(question, answer));
+        YamlConfiguration yaml = new YamlConfiguration();
+        for (int i = 0; i < entries.size(); i++) {
+            yaml.set("c" + i + ".pergunta", entries.get(i).question());
+            yaml.set("c" + i + ".resposta", entries.get(i).answer());
+        }
+        try {
+            yaml.save(file);
+        } catch (Exception e) {
+            throw new IllegalStateException("não consegui gravar " + file, e);
+        }
+    }
+
+    List<Entry> all() {
+        return List.copyOf(entries);
+    }
+
+    /** Corrections sharing at least one significant word with the question. */
+    static List<Entry> matching(List<Entry> all, String question) {
+        Set<String> asked = significantWords(question);
+        List<Entry> out = new ArrayList<>();
+        for (Entry entry : all) {
+            Set<String> known = significantWords(entry.question());
+            known.retainAll(asked);
+            if (known.size() >= 1) {
+                out.add(entry);
+            }
+        }
+        return out;
+    }
+
+    private static Set<String> significantWords(String text) {
+        Set<String> words = new HashSet<>();
+        for (String word : text.toLowerCase(Locale.ROOT).split("[^\\p{L}0-9]+")) {
+            // Short words are almost all articles and prepositions in Portuguese.
+            if (word.length() > 4) {
+                words.add(word);
+            }
+        }
+        return words;
+    }
+}

src/test/java/dev/marcospaulo/canalhandia/CorrectionsTest.java

@@ -0,0 +1,39 @@
+package dev.marcospaulo.canalhandia;
+
+import org.junit.jupiter.api.Test;
+import java.util.List;
+import static org.junit.jupiter.api.Assertions.*;
+
+class CorrectionsTest {
+
+    @Test
+    void matchesOnSharedSignificantWords() {
+        List<Corrections.Entry> all = List.of(
+                new Corrections.Entry("como pegar um camelo",
+                        "camelos são mansos, basta pôr uma sela"));
+        assertEquals(1, Corrections.matching(all, "como eu pego camelo?").size());
+    }
+
+    @Test
+    void ignoresUnrelatedCorrections() {
+        List<Corrections.Entry> all = List.of(
+                new Corrections.Entry("como pegar um camelo", "…"));
+        assertTrue(Corrections.matching(all, "onde acho diamante?").isEmpty());
+    }
+
+    @Test
+    void shortWordsDoNotCreateMatches() {
+        List<Corrections.Entry> all = List.of(
+                new Corrections.Entry("o que e um creeper", "explode"));
+        assertTrue(Corrections.matching(all, "o que e um zumbi").isEmpty());
+    }
+
+    @Test
+    void singleSharedContentWordMatches() {
+        // "camelo" is the only word > 4 chars shared with the conjugated
+        // question; the threshold is one, not two, precisely for this case.
+        List<Corrections.Entry> all = List.of(
+                new Corrections.Entry("como pegar um camelo", "ponha uma sela"));
+        assertEquals(1, Corrections.matching(all, "como eu pego camelo?").size());
+    }
+}