// Multiple spaces should be collapsed to single space
        assertTrue("Should query reading_0", queryString.contains("reading_0"));
        assertTrue("Should query reading_1", queryString.contains("reading_1"));
        // Should NOT have reading_2 (spaces collapsed)
        assertFalse("Should not have reading_2", queryString.contains("reading_2"));
    }

    @Test
    public void test_buildQuery_withNormalizer() {

                if (readings != null && readings.length > i && readings[i].length > 0) {
                    for (final String reading : readings[i]) {
                        if (!l.contains(reading)) {
                            l.add(reading);
                        }
                    }
                }

                wordsList.add(word);

        Suggester suggester = createSuggester();
        suggester.createIndexIfNothing();

        // Index some data
        String[][] readings = new String[1][];
        readings[0] = new String[] { "test" };
        SuggestItem item = new SuggestItem(new String[] { "テスト" }, readings, new String[] { "content" }, 1, 0, -1, new String[] { "tag1" },
                new String[] { SuggestConstants.DEFAULT_ROLE }, null, SuggestItem.Kind.DOCUMENT);

        String[] words = new String[] { "word1", "word2", "word3" };
        // Only provide readings for first word
        String[][] readings = new String[][] { new String[] { "reading1" } };
        String[] fields = new String[] { "content" };

        SuggestItem item =
                parser.parseSearchWords(words, readings, fields, null, null, 1, defaultReadingConverter, defaultNormalizer, analyzer, null);

### SuggestItem (Domain Entity)
Location: `src/main/java/org/codelibs/fess/suggest/entity/SuggestItem.java`

Core attributes: `text`, `timestamp`, `queryFreq`, `docFreq`, `userBoost`, `readings`, `tags`, `roles`, `languages`, `kinds`

Kind types: `DOCUMENT`, `QUERY`, `USER`

---

## Development Workflow

### Build Commands

```bash

Hello again.

////

//// tab | Info

The only hard rule for headings is that the LLM leaves the hash part inside curly brackets unchanged, which ensures that links do not break.

See section `### Headings` in the general prompt in `scripts/translate.py`.

For some language specific instructions, see e.g. section `### Headings` in `docs/de/llm-prompt.md`.

////

 * <li>{@link #badWords} - List of bad words.</li>
 * <li>{@link #parallel} - Flag for parallel processing.</li>
 * <li>{@link #readingConverter} - Reading converter.</li>
 * <li>{@link #contentsReadingConverter} - Contents reading converter.</li>
 * <li>{@link #normalizer} - Normalizer.</li>
 * <li>{@link #analyzer} - Suggest analyzer.</li>
 * <li>{@link #contentsParser} - Contents parser.</li>

     */
    public void setSuggestDetail(final boolean suggestDetail) {
        this.suggestDetail = suggestDetail;
    }

    /**
     * Sets the reading converter.
     * @param readingConverter The reading converter.
     */
    public void setReadingConverter(final ReadingConverter readingConverter) {
        this.readingConverter = readingConverter;
    }

    /**

And in many cases, reading them would be a blocking operation (that could block the event loop), because they are read from disk or from the network.

/// info

The example above is actually an exception, because the `io.BytesIO` object is already in memory, so reading it won't block anything.

But in many cases reading a file or a file-like object would block.

///

Единственное жёсткое правило для заголовков — LLM должна оставить часть хеша в фигурных скобках без изменений, чтобы ссылки не ломались.

См. раздел `### Headings` в общем промпте в `scripts/translate.py`.

Для некоторых языковых инструкций см., например, раздел `### Headings` в `docs/de/llm-prompt.md`.

////

## Термины, используемые в документации { #terms-used-in-the-docs }

//// tab | Тест

* вы
* ваш