acecode/src/utils/token_tracker.cpp at master · tmoonlight/acecode · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
#include "token_tracker.hpp"
#include <sstream>
#include <iomanip>
#include <algorithm>
#include <cstdint>

namespace acecode {

void TokenTracker::record(const TokenUsage& usage) {
    std::lock_guard<std::mutex> lk(mu_);
    // Update "last" counters (current context occupancy)
    last_prompt_tokens_ = usage.prompt_tokens;
    last_completion_tokens_ = usage.completion_tokens;
    last_total_tokens_ = usage.total_tokens;
    last_cache_read_tokens_ = usage.cache_read_tokens;
    last_cache_write_tokens_ = usage.cache_write_tokens;
    last_reasoning_tokens_ = usage.reasoning_tokens;
    last_has_data_ = usage.has_data;

    // Accumulate session totals
    session_prompt_tokens_ += usage.prompt_tokens;
    session_completion_tokens_ += usage.completion_tokens;
    session_total_tokens_ += usage.total_tokens;
    session_cache_read_tokens_ += usage.cache_read_tokens;
    session_cache_write_tokens_ += usage.cache_write_tokens;
    session_reasoning_tokens_ += usage.reasoning_tokens;
    session_has_data_ = session_has_data_ || usage.has_data;
}

void TokenTracker::record_estimate(int char_count) {
    std::lock_guard<std::mutex> lk(mu_);
    int estimated = char_count / 4;
    // For estimates, update last_prompt as well (best guess for current context)
    last_prompt_tokens_ = estimated;
    last_completion_tokens_ = 0;
    last_total_tokens_ = estimated;
    last_cache_read_tokens_ = 0;
    last_cache_write_tokens_ = 0;
    last_reasoning_tokens_ = 0;
    last_has_data_ = false;
    session_total_tokens_ += estimated;
}

int TokenTracker::last_prompt_tokens() const {
    std::lock_guard<std::mutex> lk(mu_);
    return last_prompt_tokens_;
}

int TokenTracker::last_completion_tokens() const {
    std::lock_guard<std::mutex> lk(mu_);
    return last_completion_tokens_;
}

int TokenTracker::prompt_tokens() const {
    std::lock_guard<std::mutex> lk(mu_);
    return session_prompt_tokens_;
}

int TokenTracker::completion_tokens() const {
    std::lock_guard<std::mutex> lk(mu_);
    return session_completion_tokens_;
}

int TokenTracker::total_tokens() const {
    std::lock_guard<std::mutex> lk(mu_);
    return session_total_tokens_;
}

int TokenTracker::cache_read_tokens() const {
    std::lock_guard<std::mutex> lk(mu_);
    return session_cache_read_tokens_;
}

int TokenTracker::cache_write_tokens() const {
    std::lock_guard<std::mutex> lk(mu_);
    return session_cache_write_tokens_;
}

TokenTracker::Snapshot TokenTracker::snapshot() const {
    std::lock_guard<std::mutex> lk(mu_);
    Snapshot out;
    out.last.prompt_tokens = last_prompt_tokens_;
    out.last.completion_tokens = last_completion_tokens_;
    out.last.total_tokens = last_total_tokens_;
    out.last.cache_read_tokens = last_cache_read_tokens_;
    out.last.cache_write_tokens = last_cache_write_tokens_;
    out.last.reasoning_tokens = last_reasoning_tokens_;
    out.last.has_data = last_has_data_;

    out.session.prompt_tokens = session_prompt_tokens_;
    out.session.completion_tokens = session_completion_tokens_;
    out.session.total_tokens = session_total_tokens_;
    out.session.cache_read_tokens = session_cache_read_tokens_;
    out.session.cache_write_tokens = session_cache_write_tokens_;
    out.session.reasoning_tokens = session_reasoning_tokens_;
    out.session.has_data = session_has_data_;
    return out;
}

void TokenTracker::restore(const Snapshot& snapshot) {
    restore(snapshot.last, snapshot.session);
}

void TokenTracker::restore(const TokenUsage& last, const TokenUsage& session) {
    std::lock_guard<std::mutex> lk(mu_);
    last_prompt_tokens_ = last.prompt_tokens;
    last_completion_tokens_ = last.completion_tokens;
    last_total_tokens_ = last.total_tokens;
    last_cache_read_tokens_ = last.cache_read_tokens;
    last_cache_write_tokens_ = last.cache_write_tokens;
    last_reasoning_tokens_ = last.reasoning_tokens;
    last_has_data_ = last.has_data;

    session_prompt_tokens_ = session.prompt_tokens;
    session_completion_tokens_ = session.completion_tokens;
    session_total_tokens_ = session.total_tokens;
    session_cache_read_tokens_ = session.cache_read_tokens;
    session_cache_write_tokens_ = session.cache_write_tokens;
    session_reasoning_tokens_ = session.reasoning_tokens;
    session_has_data_ = session.has_data;
}

void TokenTracker::reset() {
    std::lock_guard<std::mutex> lk(mu_);
    last_prompt_tokens_ = 0;
    last_completion_tokens_ = 0;
    last_total_tokens_ = 0;
    last_cache_read_tokens_ = 0;
    last_cache_write_tokens_ = 0;
    last_reasoning_tokens_ = 0;
    last_has_data_ = false;
    session_prompt_tokens_ = 0;
    session_completion_tokens_ = 0;
    session_total_tokens_ = 0;
    session_cache_read_tokens_ = 0;
    session_cache_write_tokens_ = 0;
    session_reasoning_tokens_ = 0;
    session_has_data_ = false;
}

std::string TokenTracker::format_tokens(int count) {
    if (count < 1000) {
        return std::to_string(count);
    }
    std::ostringstream oss;
    if (count < 100000) {
        oss << std::fixed << std::setprecision(1) << (count / 1000.0) << "k";
    } else {
        oss << (count / 1000) << "k";
    }
    return oss.str();
}

std::string TokenTracker::format_status(int context_window) const {
    std::lock_guard<std::mutex> lk(mu_);
    // Display current context occupancy (last prompt tokens), not cumulative total
    return format_tokens(last_prompt_tokens_) + "/" + format_tokens(context_window);
}

int TokenTracker::cache_hit_percent_for(int prompt_tokens, int cache_read_tokens) {
    if (prompt_tokens <= 0 || cache_read_tokens < 0) return -1;
    const auto cached = static_cast<std::int64_t>(cache_read_tokens);
    const auto prompt = static_cast<std::int64_t>(prompt_tokens);
    const auto rounded = (cached * 100 + prompt / 2) / prompt;
    return static_cast<int>(std::clamp<std::int64_t>(rounded, 0, 100));
}

int TokenTracker::last_cache_hit_percent() const {
    std::lock_guard<std::mutex> lk(mu_);
    if (!last_has_data_) return -1;
    return cache_hit_percent_for(last_prompt_tokens_, last_cache_read_tokens_);
}

int TokenTracker::cache_hit_percent() const {
    std::lock_guard<std::mutex> lk(mu_);
    if (!session_has_data_) return -1;
    return cache_hit_percent_for(session_prompt_tokens_, session_cache_read_tokens_);
}

std::string TokenTracker::format_cache_status_for(int cache_hit_percent) {
    if (cache_hit_percent < 0) return {};
    return "cache " + std::to_string(std::clamp(cache_hit_percent, 0, 100)) + "%";
}

std::string TokenTracker::format_cache_status() const {
    return format_cache_status_for(cache_hit_percent());
}

int TokenTracker::context_percent(int context_window) const {
    std::lock_guard<std::mutex> lk(mu_);
    return context_percent_for(last_prompt_tokens_, context_window);
}

int TokenTracker::context_percent_for(int prompt_tokens, int context_window) {
    if (prompt_tokens <= 0 || context_window <= 0) {
        return 0;
    }
    const auto prompt = static_cast<std::int64_t>(prompt_tokens);
    const auto context = static_cast<std::int64_t>(context_window);
    const auto rounded = (prompt * 100 + context / 2) / context;
    return static_cast<int>(std::clamp<std::int64_t>(rounded, 0, 100));
}

} // namespace acecode