defines.h revision e90b333017c68e888a5e3d351f07ea29036457d0
1e808e436cbd6f1aeadb5d61f354d03c3c50872a7satok/*
2e808e436cbd6f1aeadb5d61f354d03c3c50872a7satok**
3e808e436cbd6f1aeadb5d61f354d03c3c50872a7satok** Copyright 2010, The Android Open Source Project
4e808e436cbd6f1aeadb5d61f354d03c3c50872a7satok**
5e808e436cbd6f1aeadb5d61f354d03c3c50872a7satok** Licensed under the Apache License, Version 2.0 (the "License");
6e808e436cbd6f1aeadb5d61f354d03c3c50872a7satok** you may not use this file except in compliance with the License.
7e808e436cbd6f1aeadb5d61f354d03c3c50872a7satok** You may obtain a copy of the License at
8e808e436cbd6f1aeadb5d61f354d03c3c50872a7satok**
9e808e436cbd6f1aeadb5d61f354d03c3c50872a7satok**     http://www.apache.org/licenses/LICENSE-2.0
10e808e436cbd6f1aeadb5d61f354d03c3c50872a7satok**
11e808e436cbd6f1aeadb5d61f354d03c3c50872a7satok** Unless required by applicable law or agreed to in writing, software
12e808e436cbd6f1aeadb5d61f354d03c3c50872a7satok** distributed under the License is distributed on an "AS IS" BASIS,
13e808e436cbd6f1aeadb5d61f354d03c3c50872a7satok** WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
14e808e436cbd6f1aeadb5d61f354d03c3c50872a7satok** See the License for the specific language governing permissions and
15e808e436cbd6f1aeadb5d61f354d03c3c50872a7satok** limitations under the License.
16e808e436cbd6f1aeadb5d61f354d03c3c50872a7satok*/
17e808e436cbd6f1aeadb5d61f354d03c3c50872a7satok
18e808e436cbd6f1aeadb5d61f354d03c3c50872a7satok#ifndef LATINIME_DEFINES_H
19e808e436cbd6f1aeadb5d61f354d03c3c50872a7satok#define LATINIME_DEFINES_H
20e808e436cbd6f1aeadb5d61f354d03c3c50872a7satok
21e808e436cbd6f1aeadb5d61f354d03c3c50872a7satok#ifdef FLAG_DBG
22e808e436cbd6f1aeadb5d61f354d03c3c50872a7satok#include <cutils/log.h>
23e808e436cbd6f1aeadb5d61f354d03c3c50872a7satok#ifndef LOG_TAG
24e808e436cbd6f1aeadb5d61f354d03c3c50872a7satok#define LOG_TAG "LatinIME: "
25e808e436cbd6f1aeadb5d61f354d03c3c50872a7satok#endif
26cdbbea735f590784791f0c1fe33a514c4e864836satok#define DEBUG_DICT true
2761e2f85e3fe6c0cc266996bc9a3692ed5795fad3satok#define DEBUG_DICT_FULL false
28a3d78f606e8e764c22637299a58c27e195b4e1d3satok#define DEBUG_SHOW_FOUND_WORD DEBUG_DICT_FULL
29a3d78f606e8e764c22637299a58c27e195b4e1d3satok#define DEBUG_NODE DEBUG_DICT_FULL
30a3d78f606e8e764c22637299a58c27e195b4e1d3satok#define DEBUG_TRACE DEBUG_DICT_FULL
3161e2f85e3fe6c0cc266996bc9a3692ed5795fad3satok
3261e2f85e3fe6c0cc266996bc9a3692ed5795fad3satok// Profiler
3361e2f85e3fe6c0cc266996bc9a3692ed5795fad3satok#include <time.h>
3461e2f85e3fe6c0cc266996bc9a3692ed5795fad3satok#define PROF_BUF_SIZE 100
3561e2f85e3fe6c0cc266996bc9a3692ed5795fad3satokstatic double profile_buf[PROF_BUF_SIZE];
3661e2f85e3fe6c0cc266996bc9a3692ed5795fad3satokstatic double profile_old[PROF_BUF_SIZE];
3761e2f85e3fe6c0cc266996bc9a3692ed5795fad3satokstatic unsigned int profile_counter[PROF_BUF_SIZE];
3861e2f85e3fe6c0cc266996bc9a3692ed5795fad3satok
39e90b333017c68e888a5e3d351f07ea29036457d0Ken Wakasa#define PROF_RESET               prof_reset()
40e90b333017c68e888a5e3d351f07ea29036457d0Ken Wakasa#define PROF_COUNT(prof_buf_id)  ++profile_counter[prof_buf_id]
41e90b333017c68e888a5e3d351f07ea29036457d0Ken Wakasa#define PROF_OPEN                do { PROF_RESET; PROF_START(PROF_BUF_SIZE - 1); } while(0)
42e90b333017c68e888a5e3d351f07ea29036457d0Ken Wakasa#define PROF_START(prof_buf_id)  do { \
43e90b333017c68e888a5e3d351f07ea29036457d0Ken Wakasa        PROF_COUNT(prof_buf_id); profile_old[prof_buf_id] = (clock()); } while(0)
44e90b333017c68e888a5e3d351f07ea29036457d0Ken Wakasa#define PROF_CLOSE               do { PROF_END(PROF_BUF_SIZE - 1); PROF_OUTALL; } while(0)
45e90b333017c68e888a5e3d351f07ea29036457d0Ken Wakasa#define PROF_END(prof_buf_id)    profile_buf[prof_buf_id] += ((clock()) - profile_old[prof_buf_id])
46e90b333017c68e888a5e3d351f07ea29036457d0Ken Wakasa#define PROF_CLOCKOUT(prof_buf_id) \
47e90b333017c68e888a5e3d351f07ea29036457d0Ken Wakasa        LOGI("%s : clock is %f", __FUNCTION__, (clock() - profile_old[prof_buf_id]))
48e90b333017c68e888a5e3d351f07ea29036457d0Ken Wakasa#define PROF_OUTALL              do { LOGI("--- %s ---", __FUNCTION__); prof_out(); } while(0)
49e90b333017c68e888a5e3d351f07ea29036457d0Ken Wakasa
50e90b333017c68e888a5e3d351f07ea29036457d0Ken Wakasastatic void prof_reset(void) {
51e90b333017c68e888a5e3d351f07ea29036457d0Ken Wakasa    for (int i = 0; i < PROF_BUF_SIZE; ++i) {
5261e2f85e3fe6c0cc266996bc9a3692ed5795fad3satok        profile_buf[i] = 0;
5361e2f85e3fe6c0cc266996bc9a3692ed5795fad3satok        profile_old[i] = 0;
5461e2f85e3fe6c0cc266996bc9a3692ed5795fad3satok        profile_counter[i] = 0;
5561e2f85e3fe6c0cc266996bc9a3692ed5795fad3satok    }
5661e2f85e3fe6c0cc266996bc9a3692ed5795fad3satok}
5761e2f85e3fe6c0cc266996bc9a3692ed5795fad3satok
58e90b333017c68e888a5e3d351f07ea29036457d0Ken Wakasastatic void prof_out(void) {
5961e2f85e3fe6c0cc266996bc9a3692ed5795fad3satok    if (profile_counter[PROF_BUF_SIZE - 1] != 1) {
6061e2f85e3fe6c0cc266996bc9a3692ed5795fad3satok        LOGI("Error: You must call PROF_OPEN before PROF_CLOSE.");
6161e2f85e3fe6c0cc266996bc9a3692ed5795fad3satok    }
6261e2f85e3fe6c0cc266996bc9a3692ed5795fad3satok    LOGI("Total time is %6.3f ms.",
63e90b333017c68e888a5e3d351f07ea29036457d0Ken Wakasa            profile_buf[PROF_BUF_SIZE - 1] * 1000 / (double)CLOCKS_PER_SEC);
6461e2f85e3fe6c0cc266996bc9a3692ed5795fad3satok    double all = 0;
65e90b333017c68e888a5e3d351f07ea29036457d0Ken Wakasa    for (int i = 0; i < PROF_BUF_SIZE - 1; ++i) {
6661e2f85e3fe6c0cc266996bc9a3692ed5795fad3satok        all += profile_buf[i];
6761e2f85e3fe6c0cc266996bc9a3692ed5795fad3satok    }
68e90b333017c68e888a5e3d351f07ea29036457d0Ken Wakasa    if (all == 0) all = 1;
69e90b333017c68e888a5e3d351f07ea29036457d0Ken Wakasa    for (int i = 0; i < PROF_BUF_SIZE - 1; ++i) {
70e90b333017c68e888a5e3d351f07ea29036457d0Ken Wakasa        if (profile_buf[i] != 0) {
7161e2f85e3fe6c0cc266996bc9a3692ed5795fad3satok            LOGI("(%d): Used %4.2f%%, %8.4f ms. Called %d times.",
72e90b333017c68e888a5e3d351f07ea29036457d0Ken Wakasa                    i, (profile_buf[i] * 100 / all),
73e90b333017c68e888a5e3d351f07ea29036457d0Ken Wakasa                    profile_buf[i] * 1000 / (double)CLOCKS_PER_SEC, profile_counter[i]);
74e90b333017c68e888a5e3d351f07ea29036457d0Ken Wakasa        }
7561e2f85e3fe6c0cc266996bc9a3692ed5795fad3satok    }
7661e2f85e3fe6c0cc266996bc9a3692ed5795fad3satok}
7761e2f85e3fe6c0cc266996bc9a3692ed5795fad3satok
78e808e436cbd6f1aeadb5d61f354d03c3c50872a7satok#else // FLAG_DBG
79e90b333017c68e888a5e3d351f07ea29036457d0Ken Wakasa#define LOGE
80e808e436cbd6f1aeadb5d61f354d03c3c50872a7satok#define LOGI
81cdbbea735f590784791f0c1fe33a514c4e864836satok#define DEBUG_DICT false
82a3d78f606e8e764c22637299a58c27e195b4e1d3satok#define DEBUG_DICT_FULL false
83cdbbea735f590784791f0c1fe33a514c4e864836satok#define DEBUG_SHOW_FOUND_WORD false
84aee09dc5fadba630bbbca137cafb4871fb823099satok#define DEBUG_NODE false
85a3d78f606e8e764c22637299a58c27e195b4e1d3satok#define DEBUG_TRACE false
8661e2f85e3fe6c0cc266996bc9a3692ed5795fad3satok
8761e2f85e3fe6c0cc266996bc9a3692ed5795fad3satok#define PROF_BUF_SIZE 0
8861e2f85e3fe6c0cc266996bc9a3692ed5795fad3satok#define PROF_RESET
8961e2f85e3fe6c0cc266996bc9a3692ed5795fad3satok#define PROF_COUNT(prof_buf_id)
9061e2f85e3fe6c0cc266996bc9a3692ed5795fad3satok#define PROF_OPEN
9161e2f85e3fe6c0cc266996bc9a3692ed5795fad3satok#define PROF_START(prof_buf_id)
9261e2f85e3fe6c0cc266996bc9a3692ed5795fad3satok#define PROF_CLOSE
9361e2f85e3fe6c0cc266996bc9a3692ed5795fad3satok#define PROF_END(prof_buf_id)
9461e2f85e3fe6c0cc266996bc9a3692ed5795fad3satok#define PROF_CLOCK_OUT(prof_buf_id)
9561e2f85e3fe6c0cc266996bc9a3692ed5795fad3satok#define PROF_CLOCKOUT(prof_buf_id)
9661e2f85e3fe6c0cc266996bc9a3692ed5795fad3satok#define PROF_OUTALL
9761e2f85e3fe6c0cc266996bc9a3692ed5795fad3satok
98e808e436cbd6f1aeadb5d61f354d03c3c50872a7satok#endif // FLAG_DBG
99e808e436cbd6f1aeadb5d61f354d03c3c50872a7satok
100662fe69ba2b8513a1a6640adde917db9a13e98afsatok#ifndef U_SHORT_MAX
101662fe69ba2b8513a1a6640adde917db9a13e98afsatok#define U_SHORT_MAX 1 << 16
102662fe69ba2b8513a1a6640adde917db9a13e98afsatok#endif
103662fe69ba2b8513a1a6640adde917db9a13e98afsatok
104e90b333017c68e888a5e3d351f07ea29036457d0Ken Wakasa// Define this to use mmap() for dictionary loading.  Undefine to use malloc() instead of mmap().
105e90b333017c68e888a5e3d351f07ea29036457d0Ken Wakasa// We measured and compared performance of both, and found mmap() is fairly good in terms of
106e90b333017c68e888a5e3d351f07ea29036457d0Ken Wakasa// loading time, and acceptable even for several initial lookups which involve page faults.
107e90b333017c68e888a5e3d351f07ea29036457d0Ken Wakasa#define USE_MMAP_FOR_DICTIONARY
108e90b333017c68e888a5e3d351f07ea29036457d0Ken Wakasa
109e808e436cbd6f1aeadb5d61f354d03c3c50872a7satok// 22-bit address = ~4MB dictionary size limit, which on average would be about 200k-300k words
110e808e436cbd6f1aeadb5d61f354d03c3c50872a7satok#define ADDRESS_MASK 0x3FFFFF
111e808e436cbd6f1aeadb5d61f354d03c3c50872a7satok
112e808e436cbd6f1aeadb5d61f354d03c3c50872a7satok// The bit that decides if an address follows in the next 22 bits
113e808e436cbd6f1aeadb5d61f354d03c3c50872a7satok#define FLAG_ADDRESS_MASK 0x40
114e808e436cbd6f1aeadb5d61f354d03c3c50872a7satok// The bit that decides if this is a terminal node for a word. The node could still have children,
115e808e436cbd6f1aeadb5d61f354d03c3c50872a7satok// if the word has other endings.
116e808e436cbd6f1aeadb5d61f354d03c3c50872a7satok#define FLAG_TERMINAL_MASK 0x80
117e808e436cbd6f1aeadb5d61f354d03c3c50872a7satok
118e808e436cbd6f1aeadb5d61f354d03c3c50872a7satok#define FLAG_BIGRAM_READ 0x80
119e808e436cbd6f1aeadb5d61f354d03c3c50872a7satok#define FLAG_BIGRAM_CHILDEXIST 0x40
120e808e436cbd6f1aeadb5d61f354d03c3c50872a7satok#define FLAG_BIGRAM_CONTINUED 0x80
121e808e436cbd6f1aeadb5d61f354d03c3c50872a7satok#define FLAG_BIGRAM_FREQ 0x7F
122e808e436cbd6f1aeadb5d61f354d03c3c50872a7satok
123e808e436cbd6f1aeadb5d61f354d03c3c50872a7satok#define DICTIONARY_VERSION_MIN 200
124e808e436cbd6f1aeadb5d61f354d03c3c50872a7satok#define DICTIONARY_HEADER_SIZE 2
125e808e436cbd6f1aeadb5d61f354d03c3c50872a7satok#define NOT_VALID_WORD -99
126e808e436cbd6f1aeadb5d61f354d03c3c50872a7satok
127662fe69ba2b8513a1a6640adde917db9a13e98afsatok#define SUGGEST_WORDS_WITH_MISSING_CHARACTER true
128662fe69ba2b8513a1a6640adde917db9a13e98afsatok#define SUGGEST_WORDS_WITH_MISSING_SPACE_CHARACTER true
129662fe69ba2b8513a1a6640adde917db9a13e98afsatok#define SUGGEST_WORDS_WITH_EXCESSIVE_CHARACTER true
130a3d78f606e8e764c22637299a58c27e195b4e1d3satok#define SUGGEST_WORDS_WITH_TRANSPOSED_CHARACTERS true
131a3d78f606e8e764c22637299a58c27e195b4e1d3satok
132a3d78f606e8e764c22637299a58c27e195b4e1d3satok#define WORDS_WITH_MISSING_CHARACTER_DEMOTION_RATE 75
133a3d78f606e8e764c22637299a58c27e195b4e1d3satok#define WORDS_WITH_MISSING_SPACE_CHARACTER_DEMOTION_RATE 80
134a3d78f606e8e764c22637299a58c27e195b4e1d3satok#define WORDS_WITH_EXCESSIVE_CHARACTER_DEMOTION_RATE 75
13554fe9e0e20896f8d9813aeac7953ead9369bd4adsatok#define WORDS_WITH_EXCESSIVE_CHARACTER_OUT_OF_PROXIMITY_DEMOTION_RATE 75
136a3d78f606e8e764c22637299a58c27e195b4e1d3satok#define WORDS_WITH_TRANSPOSED_CHARACTERS_DEMOTION_RATE 60
137e808e436cbd6f1aeadb5d61f354d03c3c50872a7satok
138f5cded1c6cf0f39df13750d4f9f5ba66c1b32964satok// This should be greater than or equal to MAX_WORD_LENGTH defined in BinaryDictionary.java
139f5cded1c6cf0f39df13750d4f9f5ba66c1b32964satok// This is only used for the size of array. Not to be used in c functions.
140f5cded1c6cf0f39df13750d4f9f5ba66c1b32964satok#define MAX_WORD_LENGTH_INTERNAL 48
141715514d7dd5716c77781b97d4f6ac8eace75e8b3satok
142683192684c18d068573e5ae0c14da3839fbdbf74satok#define MAX_DEPTH_MULTIPLIER 3
143683192684c18d068573e5ae0c14da3839fbdbf74satok
14454fe9e0e20896f8d9813aeac7953ead9369bd4adsatok// Minimum suggest depth for one word for all cases except for missing space suggestions.
14554fe9e0e20896f8d9813aeac7953ead9369bd4adsatok#define MIN_SUGGEST_DEPTH 1
14654fe9e0e20896f8d9813aeac7953ead9369bd4adsatok#define MIN_USER_TYPED_LENGTH_FOR_MISSING_SPACE_SUGGESTION 3
14754fe9e0e20896f8d9813aeac7953ead9369bd4adsatok#define MIN_USER_TYPED_LENGTH_FOR_EXCESSIVE_CHARACTER_SUGGESTION 3
148662fe69ba2b8513a1a6640adde917db9a13e98afsatok
149f5cded1c6cf0f39df13750d4f9f5ba66c1b32964satok#define min(a,b) ((a)<(b)?(a):(b))
150f5cded1c6cf0f39df13750d4f9f5ba66c1b32964satok
151e808e436cbd6f1aeadb5d61f354d03c3c50872a7satok#endif // LATINIME_DEFINES_H
152