michael@0: // Copyright 2013 Google Inc. All Rights Reserved.
michael@0: //
michael@0: // Licensed under the Apache License, Version 2.0 (the "License");
michael@0: // you may not use this file except in compliance with the License.
michael@0: // You may obtain a copy of the License at
michael@0: //
michael@0: //     http://www.apache.org/licenses/LICENSE-2.0
michael@0: //
michael@0: // Unless required by applicable law or agreed to in writing, software
michael@0: // distributed under the License is distributed on an "AS IS" BASIS,
michael@0: // WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
michael@0: // See the License for the specific language governing permissions and
michael@0: // limitations under the License.
michael@0: 
michael@0: //
michael@0: // Author: dsites@google.com (Dick Sites)
michael@0: //
michael@0: 
michael@0: #ifndef I18N_ENCODINGS_COMPACT_LANG_DET_COMPACT_LANG_DET_HINT_CODE_H__
michael@0: #define I18N_ENCODINGS_COMPACT_LANG_DET_COMPACT_LANG_DET_HINT_CODE_H__
michael@0: 
michael@0: 
michael@0: #include <string>
michael@0: #include "integral_types.h"
michael@0: #include "lang_script.h"
michael@0: #include "../public/encodings.h"
michael@0: 
michael@0: namespace CLD2 {
michael@0: 
michael@0: // Packed <Language, weight>, weight in [-32..31] (powers of 2**1.6 ~=3.03)
michael@0: // Full language in bottom 10 bits, weight in top 6 bits
michael@0: typedef int16 OneCLDLangPrior;
michael@0: 
michael@0: const int kMaxOneCLDLangPrior = 14;
michael@0: typedef struct {
michael@0:   int32 n;
michael@0:   OneCLDLangPrior prior[kMaxOneCLDLangPrior];
michael@0: } CLDLangPriors;
michael@0: 
michael@0: // Reading exposed here; setting hidden in .cc
michael@0: inline int GetCLDPriorWeight(OneCLDLangPrior olp) {
michael@0:   return olp >> 10;
michael@0: }
michael@0: inline Language GetCLDPriorLang(OneCLDLangPrior olp) {
michael@0:   return static_cast<Language>(olp & 0x3ff);
michael@0: }
michael@0: 
michael@0: inline int32 GetCLDLangPriorCount(CLDLangPriors* lps) {
michael@0:   return lps->n;
michael@0: }
michael@0: 
michael@0: inline void InitCLDLangPriors(CLDLangPriors* lps) {
michael@0:   lps->n = 0;
michael@0: }
michael@0: 
michael@0: // Trim language priors to no more than max_entries, keeping largest abs weights
michael@0: void TrimCLDLangPriors(int max_entries, CLDLangPriors* lps);
michael@0: 
michael@0: // Trim language tag string to canonical form for each language
michael@0: // Input is from GetLangTagsFromHtml(), already lowercased
michael@0: std::string TrimCLDLangTagsHint(const std::string& langtags);
michael@0: 
michael@0: // Add hints to vector of langpriors
michael@0: // Input is from GetLangTagsFromHtml(), already lowercased
michael@0: void SetCLDLangTagsHint(const std::string& langtags, CLDLangPriors* langpriors);
michael@0: 
michael@0: // Add hints to vector of langpriors
michael@0: // Input is from HTTP content-language
michael@0: void SetCLDContentLangHint(const char* contentlang, CLDLangPriors* langpriors);
michael@0: 
michael@0: // Add hints to vector of langpriors
michael@0: // Input is from GetTLD(), already lowercased
michael@0: void SetCLDTLDHint(const char* tld, CLDLangPriors* langpriors);
michael@0: 
michael@0: // Add hints to vector of langpriors
michael@0: // Input is from DetectEncoding()
michael@0: void SetCLDEncodingHint(Encoding enc, CLDLangPriors* langpriors);
michael@0: 
michael@0: // Add hints to vector of langpriors
michael@0: // Input is from random source
michael@0: void SetCLDLanguageHint(Language lang, CLDLangPriors* langpriors);
michael@0: 
michael@0: // Make printable string of priors
michael@0: std::string DumpCLDLangPriors(const CLDLangPriors* langpriors);
michael@0: 
michael@0: 
michael@0: // Get language tag hints from HTML body
michael@0: // Normalize: remove spaces and make lowercase comma list
michael@0: std::string GetLangTagsFromHtml(const char* utf8_body, int32 utf8_body_len,
michael@0:                            int32 max_scan_bytes);
michael@0: 
michael@0: }       // End namespace CLD2
michael@0: 
michael@0: #endif  // I18N_ENCODINGS_COMPACT_LANG_DET_COMPACT_LANG_DET_HINT_CODE_H__
michael@0: