Skip to main content

lean_ctx/tools/
ctx_benchmark.rs

1use std::path::Path;
2
3use crate::core::compressor;
4use crate::core::entropy;
5use crate::core::quality;
6use crate::core::signatures;
7use crate::core::symbol_map::{self, SymbolMap};
8use crate::core::tokens::count_tokens;
9use crate::tools::CrpMode;
10
11pub fn handle(path: &str, crp_mode: CrpMode) -> String {
12    let content = match std::fs::read_to_string(path) {
13        Ok(c) => c,
14        Err(e) => return format!("ERROR: {e}"),
15    };
16
17    let line_count = content.lines().count();
18    let short = crate::core::protocol::shorten_path(path);
19    let ext = Path::new(path)
20        .extension()
21        .and_then(|e| e.to_str())
22        .unwrap_or("");
23
24    let raw_tokens = count_tokens(&content);
25
26    let aggressive = compressor::aggressive_compress(&content, Some(ext));
27    let aggressive_tokens = count_tokens(&aggressive);
28
29    let sigs = signatures::extract_signatures(&content, ext);
30    let sig_compact: String = sigs
31        .iter()
32        .map(super::super::core::signatures::Signature::to_compact)
33        .collect::<Vec<_>>()
34        .join("\n");
35    let sig_tokens = count_tokens(&sig_compact);
36
37    let sig_tdd: String = sigs
38        .iter()
39        .map(super::super::core::signatures::Signature::to_tdd)
40        .collect::<Vec<_>>()
41        .join("\n");
42    let sig_tdd_tokens = count_tokens(&sig_tdd);
43
44    let entropy_result = entropy::entropy_compress(&content);
45    let entropy_tokens = entropy_result.compressed_tokens;
46
47    let cache_hit = format!("F? cached 2t {line_count}L");
48    let cache_tokens = count_tokens(&cache_hit);
49
50    let mut sym = SymbolMap::new();
51    let idents = symbol_map::extract_identifiers(&content, ext);
52    for ident in &idents {
53        sym.register(ident);
54    }
55    let tdd_full = sym.apply(&content);
56    let tdd_table = sym.format_table();
57    let tdd_full_tokens = count_tokens(&tdd_full) + count_tokens(&tdd_table);
58
59    let tdd_agg = sym.apply(&aggressive);
60    let tdd_agg_tokens = count_tokens(&tdd_agg) + count_tokens(&tdd_table);
61
62    let mut rows = Vec::new();
63    rows.push(format!("Benchmark: {short} ({line_count}L)\n"));
64
65    let q_aggressive = quality_cell(&content, &aggressive, ext);
66    let q_sig_compact = quality_cell(&content, &sig_compact, ext);
67    let q_sig_tdd = quality_cell(&content, &sig_tdd, ext);
68    let q_entropy = quality_cell(&content, &entropy_result.output, ext);
69
70    if crp_mode.is_tdd() {
71        rows.push(format!(
72            "{:<28} {:>6}  {:>8}  {:>7}",
73            "Strategy", "Tokens", "Savings", "Quality"
74        ));
75        rows.push("─".repeat(57));
76        rows.push(format_row("raw", raw_tokens, raw_tokens, "—"));
77        rows.push(format_row(
78            "aggressive",
79            aggressive_tokens,
80            raw_tokens,
81            &q_aggressive,
82        ));
83        rows.push(format_row(
84            "signatures (compact)",
85            sig_tokens,
86            raw_tokens,
87            &q_sig_compact,
88        ));
89        rows.push(format_row(
90            "signatures (tdd)",
91            sig_tdd_tokens,
92            raw_tokens,
93            &q_sig_tdd,
94        ));
95        rows.push(format_row(
96            "entropy",
97            entropy_tokens,
98            raw_tokens,
99            &q_entropy,
100        ));
101        rows.push(format_row(
102            "full + §MAP (tdd)",
103            tdd_full_tokens,
104            raw_tokens,
105            "—",
106        ));
107        rows.push(format_row(
108            "aggressive + §MAP (tdd)",
109            tdd_agg_tokens,
110            raw_tokens,
111            "—",
112        ));
113        rows.push(format_row("cache hit", cache_tokens, raw_tokens, "—"));
114        rows.push("─".repeat(57));
115
116        let strategies = [
117            ("aggressive", aggressive_tokens),
118            ("signatures (compact)", sig_tokens),
119            ("signatures (tdd)", sig_tdd_tokens),
120            ("entropy", entropy_tokens),
121            ("full + §MAP", tdd_full_tokens),
122            ("aggressive + §MAP", tdd_agg_tokens),
123            ("cache hit", cache_tokens),
124        ];
125        if let Some(best) = strategies.iter().min_by_key(|(_, t)| *t) {
126            let saved = raw_tokens.saturating_sub(best.1);
127            let pct = if raw_tokens > 0 {
128                (saved as f64 / raw_tokens as f64 * 100.0).round() as usize
129            } else {
130                0
131            };
132            rows.push(format!(
133                "Best: \"{}\" saves {} tokens ({}%)",
134                best.0, saved, pct
135            ));
136        }
137
138        let tdd_extra = sig_tokens.saturating_sub(sig_tdd_tokens);
139        let tdd_pct = if sig_tokens > 0 {
140            (tdd_extra as f64 / sig_tokens as f64 * 100.0).round() as usize
141        } else {
142            0
143        };
144        rows.push(format!(
145            "TDD bonus (signatures): {tdd_extra} extra tokens saved ({tdd_pct}%)"
146        ));
147    } else {
148        rows.push(format!(
149            "{:<24} {:>6}  {:>8}  {:>7}",
150            "Strategy", "Tokens", "Savings", "Quality"
151        ));
152        rows.push("─".repeat(53));
153        rows.push(format_row("raw", raw_tokens, raw_tokens, "—"));
154        rows.push(format_row(
155            "aggressive",
156            aggressive_tokens,
157            raw_tokens,
158            &q_aggressive,
159        ));
160        rows.push(format_row(
161            "signatures (compact)",
162            sig_tokens,
163            raw_tokens,
164            &q_sig_compact,
165        ));
166        rows.push(format_row(
167            "entropy",
168            entropy_tokens,
169            raw_tokens,
170            &q_entropy,
171        ));
172        rows.push(format_row("cache hit", cache_tokens, raw_tokens, "—"));
173        rows.push("─".repeat(53));
174
175        let strategies = [
176            ("aggressive", aggressive_tokens),
177            ("signatures", sig_tokens),
178            ("entropy", entropy_tokens),
179            ("cache hit", cache_tokens),
180        ];
181        if let Some(best) = strategies.iter().min_by_key(|(_, t)| *t) {
182            let saved = raw_tokens.saturating_sub(best.1);
183            let pct = if raw_tokens > 0 {
184                (saved as f64 / raw_tokens as f64 * 100.0).round() as usize
185            } else {
186                0
187            };
188            rows.push(format!(
189                "Best: \"{}\" saves {} tokens ({}%)",
190                best.0, saved, pct
191            ));
192        }
193    }
194
195    rows.join("\n")
196}
197
198fn format_row(name: &str, tokens: usize, baseline: usize, quality: &str) -> String {
199    if tokens >= baseline {
200        format!("{name:<28} {tokens:>6}  —  {quality:>7}")
201    } else {
202        let saved = baseline - tokens;
203        let pct = (saved as f64 / baseline as f64 * 100.0).round() as usize;
204        format!("{name:<28} {tokens:>6}  -{saved} ({pct}%)  {quality:>7}")
205    }
206}
207
208fn quality_cell(original: &str, compressed: &str, ext: &str) -> String {
209    let q = quality::score(original, compressed, ext);
210    let pct = (q.composite * 100.0).round() as u32;
211    let pass = if q.passed { "✓" } else { "✗" };
212    format!("{pct:>3}%{pass}")
213}