live-caption/tests/tst_transcript.cpp

#include <QtTest>

#include "transcript.h"

// 输出解析：Qwen3-ASR 的原始输出形如 "language English<asr_text>正文"
class TstTranscript : public QObject {
    Q_OBJECT

private slots:
    void splitsLanguageAndText()
    {
        const auto r = lc::parseAsrOutput("language English<asr_text>He hoped there would be stew.");
        QCOMPARE(QString::fromStdString(r.language), QString("English"));
        QCOMPARE(QString::fromStdString(r.text), QString("He hoped there would be stew."));
    }

    void handlesChinese()
    {
        const auto r = lc::parseAsrOutput("language Chinese<asr_text>不排除购房者可以在互联网上下单。");
        QCOMPARE(QString::fromStdString(r.language), QString("Chinese"));
        QCOMPARE(QString::fromStdString(r.text), QString("不排除购房者可以在互联网上下单。"));
    }

    void withoutMarkerKeepsWholeText()
    {
        const auto r = lc::parseAsrOutput("  just text  ");
        QVERIFY(r.language.empty());
        QCOMPARE(QString::fromStdString(r.text), QString("just text"));
    }

    void trimsSurroundingWhitespaceOfBody()
    {
        const auto r = lc::parseAsrOutput("language English<asr_text>  hi there  ");
        QCOMPARE(QString::fromStdString(r.text), QString("hi there"));
    }

    void displayNameMapping()
    {
        QCOMPARE(QString::fromStdString(lc::languageDisplayName("English")), QString("英语"));
        QCOMPARE(QString::fromStdString(lc::languageDisplayName("Chinese")), QString("中文"));
        // 未收录的语言原样返回
        QCOMPARE(QString::fromStdString(lc::languageDisplayName("Klingon")), QString("Klingon"));
        QVERIFY(lc::languageDisplayName("").empty());
    }

    // 设置里的语言提示是 ISO 码，界面要把它显示成中文名。
    // 覆盖注册表里出现过的所有语言码，避免加了模型却漏了显示名。
    void isoCodeDisplayName()
    {
        const struct { const char * code; const char * display; } kCases[] = {
            { "zh", "中文" },   { "en", "英语" },     { "ja", "日语" },
            { "ko", "韩语" },   { "fr", "法语" },     { "de", "德语" },
            { "es", "西班牙语" },{ "ru", "俄语" },    { "ar", "阿拉伯语" },
            { "hi", "印地语" }, { "vi", "越南语" },   { "it", "意大利语" },
            { "pt", "葡萄牙语" },{ "nl", "荷兰语" },  { "tr", "土耳其语" },
            { "uk", "乌克兰语" },{ "auto", "自动检测" },
        };
        for (const auto & c : kCases) {
            QCOMPARE(QString::fromStdString(lc::languageDisplayNameFromCode(c.code)),
                     QString::fromUtf8(c.display));
        }
        // 未知码原样返回；空串返回空
        QCOMPARE(QString::fromStdString(lc::languageDisplayNameFromCode("xx")), QString("xx"));
        QVERIFY(lc::languageDisplayNameFromCode("").empty());
    }

    // 注册表里每个模型声明的语言都必须能显示成中文名，
    // 否则设置里的「识别语言」下拉会出现英文码。
    void registryLanguagesAllHaveDisplayNames()
    {
        for (const char * code : { "zh", "en", "ja", "ko", "fr", "de", "es", "ru", "ar",
                                   "hi", "vi", "it", "pt", "nl", "tr", "uk" }) {
            const QString display =
                QString::fromStdString(lc::languageDisplayNameFromCode(code));
            QVERIFY2(display != QString::fromLatin1(code),
                     qPrintable(QStringLiteral("语言码没有显示名: %1").arg(code)));
        }
    }
};

QTEST_MAIN(TstTranscript)
#include "tst_transcript.moc"