PageRenderTime 20ms CodeModel.GetById 13ms app.highlight 5ms RepoModel.GetById 1ms app.codeStats 0ms

/src/im/gpinyin/include/searchutility.h

http://ftk.googlecode.com/
C++ Header | 142 lines | 54 code | 26 blank | 62 comment | 0 complexity | 20abf7dc435bd7e10f222bed413022c3 MD5 | raw file
  1/*
  2 * Copyright (C) 2009 The Android Open Source Project
  3 *
  4 * Licensed under the Apache License, Version 2.0 (the "License");
  5 * you may not use this file except in compliance with the License.
  6 * You may obtain a copy of the License at
  7 *
  8 *      http://www.apache.org/licenses/LICENSE-2.0
  9 *
 10 * Unless required by applicable law or agreed to in writing, software
 11 * distributed under the License is distributed on an "AS IS" BASIS,
 12 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 13 * See the License for the specific language governing permissions and
 14 * limitations under the License.
 15 */
 16
 17#ifndef PINYINIME_ANDPY_INCLUDE_SEARCHCOMMON_H__
 18#define PINYINIME_ANDPY_INCLUDE_SEARCHCOMMON_H__
 19
 20#include <stdlib.h>
 21#include "./spellingtrie.h"
 22
 23namespace ime_pinyin {
 24
 25// Type used to identify the size of a pool, such as id pool, etc.
 26typedef uint16 PoolPosType;
 27
 28// Type used to identify a parsing mile stone in an atom dictionary.
 29typedef uint16 MileStoneHandle;
 30
 31// Type used to express a lemma and its probability score.
 32typedef struct {
 33  unsigned id:(kLemmaIdSize * 8);
 34  unsigned lma_len:4;
 35  uint16 psb;  // The score, the lower psb, the higher possibility.
 36  // For single character items, we may also need Hanzi.
 37  // For multiple characer items, ignore it.
 38  char16 hanzi;
 39} LmaPsbItem, *PLmaPsbItem;
 40
 41// LmaPsbItem extended with string.
 42typedef struct {
 43  LmaPsbItem lpi;
 44  char16 str[kMaxLemmaSize + 1];
 45} LmaPsbStrItem, *PLmaPsbStrItem;
 46
 47
 48typedef struct {
 49  float psb;
 50  char16 pre_hzs[kMaxPredictSize];
 51  uint16 his_len;  // The length of the history used to do the prediction.
 52} NPredictItem, *PNPredictItem;
 53
 54// Parameter structure used to extend in a dictionary. All dictionaries
 55// receives the same DictExtPara and a dictionary specific MileStoneHandle for
 56// extending.
 57//
 58// When the user inputs a new character, AtomDictBase::extend_dict() will be
 59// called at least once for each dictionary.
 60//
 61// For example, when the user inputs "wm", extend_dict() will be called twice,
 62// and the DictExtPara parameter are as follows respectively:
 63// 1. splids = {w, m}; splids_extended = 1; ext_len = 1; step_no = 1;
 64// splid_end_split = false; id_start = wa(the first id start with 'w');
 65// id_num = number of ids starting with 'w'.
 66// 2. splids = {m}; splids_extended = 0; ext_len = 1; step_no = 1;
 67// splid_end_split = false; id_start = wa; id_num = number of ids starting with
 68// 'w'.
 69//
 70// For string "women", one of the cases of the DictExtPara parameter is:
 71// splids = {wo, men}, splids_extended = 1, ext_len = 3 (length of "men"),
 72// step_no = 4; splid_end_split = false; id_start = men, id_num = 1.
 73//
 74typedef struct {
 75  // Spelling ids for extending, there are splids_extended + 1 ids in the
 76  // buffer.
 77  // For a normal lemma, there can only be kMaxLemmaSize spelling ids in max,
 78  // but for a composing phrase, there can kMaxSearchSteps spelling ids.
 79  uint16 splids[kMaxSearchSteps];
 80
 81  // Number of ids that have been used before. splids[splids_extended] is the
 82  // newly added id for the current extension.
 83  uint16 splids_extended;
 84
 85  // The step span of the extension. It is also the size of the string for
 86  // the newly added spelling id.
 87  uint16 ext_len;
 88
 89  // The step number for the current extension. It is also the ending position
 90  // in the input Pinyin string for the substring of spelling ids in splids[].
 91  // For example, when the user inputs "women", step_no = 4.
 92  // This parameter may useful to manage the MileStoneHandle list for each
 93  // step. When the user deletes a character from the string, MileStoneHandle
 94  // objects for the the steps after that character should be reset; when the
 95  // user begins a new string, all MileStoneHandle objects should be reset.
 96  uint16 step_no;
 97
 98  // Indicate whether the newly added spelling ends with a splitting character
 99  bool splid_end_split;
100
101  // If the newly added id is a half id, id_start is the first id of the
102  // corresponding full ids; if the newly added id is a full id, id_start is
103  // that id.
104  uint16 id_start;
105
106  // If the newly added id is a half id, id_num is the number of corresponding
107  // ids; if it is a full id, id_num == 1.
108  uint16 id_num;
109}DictExtPara, *PDictExtPara;
110
111bool is_system_lemma(LemmaIdType lma_id);
112bool is_user_lemma(LemmaIdType lma_id);
113bool is_composing_lemma(LemmaIdType lma_id);
114
115int cmp_lpi_with_psb(const void *p1, const void *p2);
116int cmp_lpi_with_unified_psb(const void *p1, const void *p2);
117int cmp_lpi_with_id(const void *p1, const void *p2);
118int cmp_lpi_with_hanzi(const void *p1, const void *p2);
119
120int cmp_lpsi_with_str(const void *p1, const void *p2);
121
122int cmp_hanzis_1(const void *p1, const void *p2);
123int cmp_hanzis_2(const void *p1, const void *p2);
124int cmp_hanzis_3(const void *p1, const void *p2);
125int cmp_hanzis_4(const void *p1, const void *p2);
126int cmp_hanzis_5(const void *p1, const void *p2);
127int cmp_hanzis_6(const void *p1, const void *p2);
128int cmp_hanzis_7(const void *p1, const void *p2);
129int cmp_hanzis_8(const void *p1, const void *p2);
130
131int cmp_npre_by_score(const void *p1, const void *p2);
132int cmp_npre_by_hislen_score(const void *p1, const void *p2);
133int cmp_npre_by_hanzi_score(const void *p1, const void *p2);
134
135
136unsigned remove_duplicate_npre(NPredictItem *npre_items, unsigned npre_num);
137
138unsigned align_to_unsigned(unsigned size);
139
140}  // namespace
141
142#endif  // PINYINIME_ANDPY_INCLUDE_SEARCHCOMMON_H__