third_party/cld/encodings/compact_lang_det/win/cld_unicodetext.h

   1 // Copyright (c) 2009 The Chromium Authors. All rights reserved.
   2 // Use of this source code is governed by a BSD-style license that can be
   3 // found in the LICENSE file.
   4
   5 #ifndef ENCODINGS_COMPACT_LANG_DET_WIN_CLD_UNICODETEXT_H_
   6 #define ENCODINGS_COMPACT_LANG_DET_WIN_CLD_UNICODETEXT_H_
   7
   8 #include "languages/public/languages.h"
   9 #include "unicode/utypes.h"
  10
  11 namespace CompactLangDet {
  12   struct DetectionTables;
  13 }  // namespace CompactLangDet
  14
  15 // Detects a language of the UTF-16 encoded zero-terminated text.
  16 // [in] detection_tables - internal CLD data tables (see compact_lang_det.h).
  17 //     Can be NULL, in this case CLD will fall back to builtin static tables.
  18 // [in] text - UTF-16 encoded text to detect a language of.
  19 // [in] is_plain_text - true if plain text, false otherwise (e.g. HTML).
  20 // [out] is_reliable - true, if returned language was detected reliably.
  21 //     See compact_lang_det.h for details.
  22 // [out] num_languages - set to the number of languages detected on the page.
  23 //     Language counts only if it's detected in more than 20% of the text.
  24 // [out, optional] error_code - set to 0 in case of success, Windows
  25 //     GetLastError() code otherwise.  Pass NULL, if not interested in errors.
  26 // See encodings/compact_lang_det/compact_lang_det.h,
  27 //     CompactLangDet::DetectLanguage() description for other input parameters
  28 //     description.
  29 // Returns: Language enum.
  30 //     Returns NUM_LANGUAGES in case of any error.
  31 //     See googleclient/languages/internal/languages.cc
  32 //     for details.
  33 Language DetectLanguageOfUnicodeText(
  34     const CompactLangDet::DetectionTables* detection_tables,
  35     const UChar* text, bool is_plain_text,
  36     bool* is_reliable, int* num_languages,
  37     int* error_code, int* text_bytes);
  38
  39
  40 #endif  // ENCODINGS_COMPACT_LANG_DET_WIN_CLD_UNICODETEXT_H_