view release on metacpan or  search on metacpan

xsp/CLD2.xsp  view on Meta::CPAN

#include <compact_lang_det.h>
#include "myinit.h"
#include "myconversions.h"

%module{Lingua::Identify::CLD2};
%package{Lingua::Identify::CLD2};

%{

void
ExtDetectLanguage(buffer, is_plain_text, flags, content_language_hint, tld_hint, encoding_hint, language_hint, returnVectors)
    SV *buffer
    bool is_plain_text
    int flags
    char* content_language_hint
    char* tld_hint
    int encoding_hint
    SV* language_hint
    bool returnVectors
  PPCODE:
    char *buffer_str;
    STRLEN buffer_length;
    buffer_str = SvPVutf8(buffer, buffer_length);

    CLDHints cld_hints_struct;
    cld_hints_struct.content_language_hint = content_language_hint && content_language_hint[0] ? content_language_hint : NULL;
    cld_hints_struct.tld_hint = tld_hint && tld_hint[0] ? tld_hint : NULL;
    cld_hints_struct.encoding_hint = encoding_hint;
    cld_hints_struct.language_hint = scalar_to_language(language_hint);

    Language language3[3] = {UNKNOWN_LANGUAGE, UNKNOWN_LANGUAGE, UNKNOWN_LANGUAGE};
    int percent3[3] = {0, 0, 0};
    double normalized_score3[3] = {0.0, 0.0, 0.0};
    ResultChunkVector resultchunkvector;
    ResultChunkVector* p_resultchunkvector = returnVectors ? &resultchunkvector : 0;
    int text_bytes = 0;
    bool is_reliable = 0;
    int valid_prefix_bytes = 0;

    Language language =
      ExtDetectLanguageSummaryCheckUTF8(buffer_str, (int)buffer_length, is_plain_text, &cld_hints_struct, flags,
                                        language3, percent3, normalized_score3, p_resultchunkvector,
                                        &text_bytes, &is_reliable, &valid_prefix_bytes);

    if (valid_prefix_bytes < buffer_length) {
        croak("input contains invalid UTF-8 around byte %d of %d", valid_prefix_bytes, buffer_length);
    }

    HV *hv = newHV();
    SV *rv = sv_2mortal(newRV_noinc((SV *)hv));

    hv_stores(hv, "language_name", newSVpv(CLD2::LanguageName(language), 0));
    hv_stores(hv, "language_code", newSVpv(CLD2::LanguageCode(language), 0));

    if (returnVectors) {
        AV *reschunk_av = resultchunk_vector_to_array(aTHX_ resultchunkvector);
        hv_stores(hv, "resultchunkvector", newRV_noinc((SV *)reschunk_av));
    }

    hv_stores(hv, "text_bytes", newSViv(text_bytes));
    hv_stores(hv, "is_reliable", newSViv((int)is_reliable));

    hv_stores(hv, "languages", newRV_noinc((SV*)languages_to_array(language3, percent3, normalized_score3)));

    XPUSHs(rv);


char *
LanguageName(int language_id)
  ALIAS:
    LanguageName = 0
    LanguageCode = 1
    LanguageDeclaredName = 2
  CODE:
    switch (ix) {
    case 0:
      RETVAL = (char *)CLD2::LanguageName((CLD2::Language)language_id);
      break;
    case 1:
      RETVAL = (char *)CLD2::LanguageCode((CLD2::Language)language_id);