static const char *enca_guess(struct mp_log *log, bstr buf, const char *language)
{
if (!language || !language[0])
language = "__"; // neutral language
const char *detected_cp = NULL;
EncaAnalyser analyser = enca_analyser_alloc(language);
if (analyser) {
enca_set_termination_strictness(analyser, 0);
EncaEncoding enc = enca_analyse_const(analyser, buf.start, buf.len);
const char *tmp = enca_charset_name(enc.charset, ENCA_NAME_STYLE_ICONV);
if (tmp && enc.charset != ENCA_CS_UNKNOWN)
detected_cp = tmp;
enca_analyser_free(analyser);
} else {
mp_err(log, "ENCA doesn't know language '%s'\n", language);
size_t langcnt;
const char **languages = enca_get_languages(&langcnt);
mp_err(log, "ENCA supported languages:");
for (int i = 0; i < langcnt; i++)
mp_err(log, " %s", languages[i]);
mp_err(log, "\n");
free(languages);
}
return detected_cp;
}
static const char *enca_guess(struct mp_log *log, bstr buf, const char *language)
{
// Do our own UTF-8 detection, because ENCA seems to get it wrong sometimes
// (suggested by divVerent). Explicitly allow cut-off UTF-8.
if (bstr_validate_utf8(buf) > -8)
return "UTF-8";
if (!language || !language[0])
language = "__"; // neutral language
const char *detected_cp = NULL;
EncaAnalyser analyser = enca_analyser_alloc(language);
if (analyser) {
enca_set_termination_strictness(analyser, 0);
EncaEncoding enc = enca_analyse_const(analyser, buf.start, buf.len);
const char *tmp = enca_charset_name(enc.charset, ENCA_NAME_STYLE_ICONV);
if (tmp && enc.charset != ENCA_CS_UNKNOWN)
detected_cp = tmp;
enca_analyser_free(analyser);
} else {
mp_err(log, "ENCA doesn't know language '%s'\n", language);
size_t langcnt;
const char **languages = enca_get_languages(&langcnt);
mp_err(log, "ENCA supported languages:");
for (int i = 0; i < langcnt; i++)
mp_err(log, " %s", languages[i]);
mp_err(log, "\n");
free(languages);
}
return detected_cp;
}
SettingsDialog::SettingsDialog(QWidget *parent) : QDialog(parent)
{
m_ui.setupUi(this);
setAttribute(Qt::WA_DeleteOnClose);
findCodecs();
foreach (QTextCodec *codec, codecs)
m_ui.icyEncodingComboBox->addItem(codec->name());
#ifdef WITH_ENCA
size_t n = 0;
const char **langs = enca_get_languages(&n);
for (size_t i = 0; i < n; ++i)
m_ui.encaAnalyserComboBox->addItem(langs[i]);
#endif
QSettings settings(Qmmp::configFile(), QSettings::IniFormat);
settings.beginGroup("HTTP");
int pos = m_ui.icyEncodingComboBox->findText(settings.value("icy_encoding","UTF-8").toString());
m_ui.icyEncodingComboBox->setCurrentIndex(pos);
m_ui.bufferSizeSpinBox->setValue(settings.value("buffer_size",384).toInt());
m_ui.userAgentCheckBox->setChecked(settings.value("override_user_agent",false).toBool());
m_ui.userAgentLineEdit->setText(settings.value("user_agent").toString());
#ifdef WITH_ENCA
m_ui.autoCharsetCheckBox->setChecked(settings.value("use_enca", false).toBool());
pos = m_ui.encaAnalyserComboBox->findText(settings.value("enca_lang", langs[n-1]).toString());
m_ui.encaAnalyserComboBox->setCurrentIndex(pos);
#else
m_ui.autoCharsetCheckBox->setEnabled(false);
#endif
settings.endGroup();
}
SettingsDialog::SettingsDialog(QWidget *parent)
: QDialog(parent)
{
ui.setupUi(this);
setAttribute(Qt::WA_DeleteOnClose);
findCodecs();
foreach (QTextCodec *codec, codecs)
ui.cueEncComboBox->addItem(codec->name());
#ifdef WITH_ENCA
size_t n = 0;
const char **langs = enca_get_languages(&n);
for (size_t i = 0; i < n; ++i)
ui.encaAnalyserComboBox->addItem(langs[i]);
#endif
QSettings settings(Qmmp::configFile(), QSettings::IniFormat);
settings.beginGroup("CUE");
int pos = ui.cueEncComboBox->findText(settings.value("encoding","UTF-8").toString());
ui.cueEncComboBox->setCurrentIndex(pos);
#ifdef WITH_ENCA
ui.autoCharsetCheckBox->setChecked(settings.value("use_enca", false).toBool());
pos = ui.encaAnalyserComboBox->findText(settings.value("enca_lang", langs[n-1]).toString());
ui.encaAnalyserComboBox->setCurrentIndex(pos);
#else
ui.autoCharsetCheckBox->setEnabled(false);
#endif
ui.dirtyCueCheckBox->setChecked(settings.value("dirty_cue", false).toBool());
settings.endGroup();
}
void *ass_guess_buffer_cp(ASS_Library *library, unsigned char *buffer,
int buflen, char *preferred_language,
char *fallback)
{
const char **languages;
size_t langcnt;
EncaAnalyser analyser;
EncaEncoding encoding;
char *detected_sub_cp = NULL;
int i;
languages = enca_get_languages(&langcnt);
ass_msg(library, MSGL_V, "ENCA supported languages");
for (i = 0; i < langcnt; i++) {
ass_msg(library, MSGL_V, "lang %s", languages[i]);
}
for (i = 0; i < langcnt; i++) {
const char *tmp;
if (strcasecmp(languages[i], preferred_language) != 0)
continue;
analyser = enca_analyser_alloc(languages[i]);
encoding = enca_analyse_const(analyser, buffer, buflen);
tmp = enca_charset_name(encoding.charset, ENCA_NAME_STYLE_ICONV);
if (tmp && encoding.charset != ENCA_CS_UNKNOWN) {
detected_sub_cp = strdup(tmp);
ass_msg(library, MSGL_INFO, "ENCA detected charset: %s", tmp);
}
enca_analyser_free(analyser);
}
free(languages);
if (!detected_sub_cp) {
detected_sub_cp = strdup(fallback);
ass_msg(library, MSGL_INFO,
"ENCA detection failed: fallback to %s", fallback);
}
return detected_sub_cp;
}
