Span<const uint8_t> Text::get_text() const
{
if (!_text_record) {
return Span<const uint8_t>();
}
size_t language_code_size = get_language_code().size();
return make_const_Span(
_text_record + header_size + language_code_size,
_text_record_size - header_size - language_code_size
);
}
dhvani_Languages detect_language(unsigned short *word, int size)
{
unsigned short unicode;
int i = 0;
/*check each letters from left to right */
while (i < size) {
unicode = word[i];
if (unicode >= 0x0D00 && unicode <= 0x0D6F) {
return MALAYALAM;
}
if (unicode >= 0x0901 && unicode <= 0x097D) {
/* Marathi mr_IN shares the code points of devangari with Hindi */
if (getenv("LANG") != NULL
&& strcmp(getenv("LANG"), "mr_IN") == 0) {
return MARATHI; /*giving priority to LANG environment variable */
} else {
return HINDI;
}
}
if (unicode >= 0x0C82 && unicode <= 0x0CEF) {
return KANNADA;
}
if (unicode >= 0x0B01 && unicode <= 0x0B71) {
return ORIYA;
}
if (unicode >= 0x0A81 && unicode <= 0x0AF1) {
return GUJARATI;
}
if (unicode >= 0x0C01 && unicode <= 0x0C6F) {
return TELUGU;
}
if (unicode >= 0x0A01 && unicode <= 0x0A74) {
return PANJABI;
}
if (unicode >= 0x0981 && unicode <= 0x09FA) {
return BENGALI;
}
if (unicode >= 0x0B82 && unicode <= 0x0BFA) {
return TAMIL;
}
if (unicode >= 0x060B && unicode <= 0xFEFC) {
return PASHTO;
}
i++;
}
/* Another attempt to see whether the LANG env variable is a supported language */
if (getenv("LANG") != NULL) {
return get_language_code(getenv("LANG"));
} else {
return -1; /*sigh... unknown language... */
}
}
