CREATE TABLE "wanikani.subjects" (
  "id"          INT NOT NULL,
  "updatedAt"   TIMESTAMPTZ NOT NULL,
  "object"      TEXT NOT NULL,
  "url"         TEXT NOT NULL,
  "data"        JSONB NOT NULL,
    -- slug: string
    -- characters?: string
    -- character_images?: { url: string; content_type: string; metadata: any }[]
    -- amalgamation_subject_ids: int[]
    -- visually_similar_subject_ids?: int[]
    -- component_subject_ids?: int[]
    -- meanings: { meaning: string; accepted_answer: boolean }[]
    -- readings: { reading: string; type: string; accepted_answer: boolean; primary: boolean }[]
    -- context_sentences?: { en: string; ja: string }[]
  PRIMARY KEY ("id")
);

CREATE INDEX "idx_wanikani.subjects_object" ON "wanikani.subjects" ("object");

CREATE INDEX "idx_wanikani.subjects_data-characters" ON "wanikani.subjects" (("data" ->> 'characters'));
CREATE INDEX "idx_wanikani.subjects_data-level" ON "wanikani.subjects" ((("data" -> 'level')::int));

CREATE INDEX "idx_wanikani.subjects_data" ON "wanikani.subjects"
  USING pgroonga("data")
  WITH (
    plugins='token_filters/stem',
    token_filters='TokenFilterStem'
  );
CREATE INDEX "idx_wanikani.subjects_data_kana" ON "wanikani.subjects"
  USING pgroonga((identity("data")))
  WITH (
    tokenizer='TokenMecab',
    normalizer='NormalizerNFKC100("unify_kana", true)'
  );

CREATE MATERIALIZED VIEW "wanikani.subjects.sentences" AS (
  SELECT
    (s ->> 'ja') "ja",
    (s ->> 'en') "en"
  FROM "wanikani.subjects", jsonb_array_elements("data" -> 'context_sentences') "s"
  WHERE s ->> 'ja' > ''
);

CREATE INDEX "idx_wanikani.subjects.sentences_ja" ON "wanikani.subjects.sentences"
  USING pgroonga("ja")
  WITH (
    tokenizer='TokenMecab',
    normalizer='NormalizerNFKC100("unify_kana", true)'
  );

CREATE INDEX "idx_wanikani.subjects.sentences_ja_kana" ON "wanikani.subjects.sentences"
  USING pgroonga((identity("ja")))
  WITH (
    tokenizer='TokenMecab("use_reading", true)',
    normalizer='NormalizerNFKC100("unify_kana", true)'
  );

CREATE INDEX "idx_wanikani.subjects.sentences_en" ON "wanikani.subjects.sentences"
  USING pgroonga("en")
  WITH (
    plugins='token_filters/stem',
    token_filters='TokenFilterStem'
  );