verifiable-thinking-mcp 0.4.0
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/LICENSE +21 -0
- package/README.md +339 -0
- package/package.json +75 -0
- package/src/index.ts +38 -0
- package/src/lib/cache.ts +246 -0
- package/src/lib/compression.ts +804 -0
- package/src/lib/compute/cache.ts +86 -0
- package/src/lib/compute/classifier.ts +555 -0
- package/src/lib/compute/confidence.ts +79 -0
- package/src/lib/compute/context.ts +154 -0
- package/src/lib/compute/extract.ts +200 -0
- package/src/lib/compute/filter.ts +224 -0
- package/src/lib/compute/index.ts +171 -0
- package/src/lib/compute/math.ts +247 -0
- package/src/lib/compute/patterns.ts +564 -0
- package/src/lib/compute/registry.ts +145 -0
- package/src/lib/compute/solvers/arithmetic.ts +65 -0
- package/src/lib/compute/solvers/calculus.ts +249 -0
- package/src/lib/compute/solvers/derivation-core.ts +371 -0
- package/src/lib/compute/solvers/derivation-latex.ts +160 -0
- package/src/lib/compute/solvers/derivation-mistakes.ts +1046 -0
- package/src/lib/compute/solvers/derivation-simplify.ts +451 -0
- package/src/lib/compute/solvers/derivation-transform.ts +620 -0
- package/src/lib/compute/solvers/derivation.ts +67 -0
- package/src/lib/compute/solvers/facts.ts +120 -0
- package/src/lib/compute/solvers/formula.ts +728 -0
- package/src/lib/compute/solvers/index.ts +36 -0
- package/src/lib/compute/solvers/logic.ts +422 -0
- package/src/lib/compute/solvers/probability.ts +307 -0
- package/src/lib/compute/solvers/statistics.ts +262 -0
- package/src/lib/compute/solvers/word-problems.ts +408 -0
- package/src/lib/compute/types.ts +107 -0
- package/src/lib/concepts.ts +111 -0
- package/src/lib/domain.ts +731 -0
- package/src/lib/extraction.ts +912 -0
- package/src/lib/index.ts +122 -0
- package/src/lib/judge.ts +260 -0
- package/src/lib/math/ast.ts +842 -0
- package/src/lib/math/index.ts +8 -0
- package/src/lib/math/operators.ts +171 -0
- package/src/lib/math/tokenizer.ts +477 -0
- package/src/lib/patterns.ts +200 -0
- package/src/lib/session.ts +825 -0
- package/src/lib/think/challenge.ts +323 -0
- package/src/lib/think/complexity.ts +504 -0
- package/src/lib/think/confidence-drift.ts +507 -0
- package/src/lib/think/consistency.ts +347 -0
- package/src/lib/think/guidance.ts +188 -0
- package/src/lib/think/helpers.ts +568 -0
- package/src/lib/think/hypothesis.ts +216 -0
- package/src/lib/think/index.ts +127 -0
- package/src/lib/think/prompts.ts +262 -0
- package/src/lib/think/route.ts +358 -0
- package/src/lib/think/schema.ts +98 -0
- package/src/lib/think/scratchpad-schema.ts +662 -0
- package/src/lib/think/spot-check.ts +961 -0
- package/src/lib/think/types.ts +93 -0
- package/src/lib/think/verification.ts +260 -0
- package/src/lib/tokens.ts +177 -0
- package/src/lib/verification.ts +620 -0
- package/src/prompts/index.ts +10 -0
- package/src/prompts/templates.ts +336 -0
- package/src/resources/index.ts +8 -0
- package/src/resources/sessions.ts +196 -0
- package/src/tools/compress.ts +138 -0
- package/src/tools/index.ts +5 -0
- package/src/tools/scratchpad.ts +2659 -0
- package/src/tools/sessions.ts +144 -0
|
@@ -0,0 +1,564 @@
|
|
|
1
|
+
/**
|
|
2
|
+
* Pre-compiled regex patterns for the compute module
|
|
3
|
+
*
|
|
4
|
+
* All patterns are compiled once at module load time, not per-call.
|
|
5
|
+
* This saves ~0.05ms per unique pattern on first call.
|
|
6
|
+
*/
|
|
7
|
+
|
|
8
|
+
import type { NegativeSignal, WeightedSignal, WordProblemMatch } from "./types.ts";
|
|
9
|
+
|
|
10
|
+
// =============================================================================
|
|
11
|
+
// GUARDS: Cheap character-based pre-checks
|
|
12
|
+
// =============================================================================
|
|
13
|
+
|
|
14
|
+
export const GUARDS = {
|
|
15
|
+
hasDigit: (t: string) => /\d/.test(t),
|
|
16
|
+
hasX: (t: string) => /x/i.test(t),
|
|
17
|
+
hasPercent: (t: string) => t.includes("%"),
|
|
18
|
+
hasCaret: (t: string) => t.includes("^") || t.includes("**"),
|
|
19
|
+
hasBracket: (t: string) => t.includes("["),
|
|
20
|
+
hasDollar: (t: string) => t.includes("$"),
|
|
21
|
+
hasExclaim: (t: string) => t.includes("!"),
|
|
22
|
+
} as const;
|
|
23
|
+
|
|
24
|
+
// =============================================================================
|
|
25
|
+
// ARITHMETIC PATTERNS
|
|
26
|
+
// =============================================================================
|
|
27
|
+
|
|
28
|
+
export const ARITHMETIC = {
|
|
29
|
+
whatIs: /what\s+is\s+([\d\s+\-*/().]+)/i,
|
|
30
|
+
calculate: /calculate[:\s]+\s*([\d\s+\-*/().]+)/i,
|
|
31
|
+
compute: /compute[:\s]+\s*([\d\s+\-*/().]+)/i,
|
|
32
|
+
evaluate: /evaluate[:\s]+\s*([\d\s+\-*/().]+)/i,
|
|
33
|
+
equalsQuestion: /([\d\s+\-*/().]+)\s*=\s*\?/,
|
|
34
|
+
bareExpression: /^([\d\s+\-*/().]+)$/,
|
|
35
|
+
validChars: /^[\d+\-*/().]+$/,
|
|
36
|
+
invalidPatterns: /\(\)|\+\+|--|\*\*|\/\/|\+-|-\+|\*\/|\/\*/,
|
|
37
|
+
} as const;
|
|
38
|
+
|
|
39
|
+
// =============================================================================
|
|
40
|
+
// FORMULA PATTERNS - Tier 1 (Ultra-fast)
|
|
41
|
+
// =============================================================================
|
|
42
|
+
|
|
43
|
+
export const TIER1 = {
|
|
44
|
+
percentage: /(\d+(?:\.\d+)?)\s*%\s*(?:of\s+)?(\d+(?:\.\d+)?)/i,
|
|
45
|
+
factorial: /(\d+)!|factorial\s*(?:of\s*)?(\d+)/i,
|
|
46
|
+
moduloBasic: /(\d+)\s*(?:mod|modulo)\s*(\d+)/i,
|
|
47
|
+
moduloRemainder: /remainder.*?(\d+).*?(?:divided\s+by|\/)\s*(\d+)/i,
|
|
48
|
+
moduloLastDigitGuard: /\^\s*\d+\s*mod\s*10/i,
|
|
49
|
+
prime: /is\s+(\d+)\s+(?:a\s+)?prime/i,
|
|
50
|
+
fibonacci: /(\d+)(?:th|st|nd|rd)\s+fibonacci/i,
|
|
51
|
+
} as const;
|
|
52
|
+
|
|
53
|
+
// =============================================================================
|
|
54
|
+
// FORMULA PATTERNS - Tier 2 (Fast)
|
|
55
|
+
// =============================================================================
|
|
56
|
+
|
|
57
|
+
export const TIER2 = {
|
|
58
|
+
sqrt: /(?:\u221A|sqrt\s*\(?\s*|square\s+root\s+(?:of\s+)?)(\d+(?:\.\d+)?)/i,
|
|
59
|
+
power: /(\d+(?:\.\d+)?)\s*(?:\^|\*\*|to\s+the\s+(?:power\s+(?:of\s+)?)?)\s*(\d+(?:\.\d+)?)/i,
|
|
60
|
+
powerLastDigitGuard: /last\s+digit/i,
|
|
61
|
+
powerModGuard: /\^\s*\d+\s*mod/i,
|
|
62
|
+
gcd: /(?:gcd|greatest\s+common\s+divisor).*?(\d+).*?(\d+)/i,
|
|
63
|
+
lcm: /(?:lcm|least\s+common\s+multiple).*?(\d+).*?(\d+)/i,
|
|
64
|
+
} as const;
|
|
65
|
+
|
|
66
|
+
// =============================================================================
|
|
67
|
+
// FORMULA PATTERNS - Tier 3 (Medium)
|
|
68
|
+
// =============================================================================
|
|
69
|
+
|
|
70
|
+
export const TIER3 = {
|
|
71
|
+
logBase10: /log[₁1]?[₀0]?\s*\(?\s*(\d+)\s*\)?/gi,
|
|
72
|
+
logNatural: /ln\s*\(?\s*(\d+(?:\.\d+)?)\s*\)?/i,
|
|
73
|
+
quadratic: /(\d*)x[\u00B22^2]\s*([+-])\s*(\d+)x\s*([+-])\s*(\d+)\s*=\s*0/i,
|
|
74
|
+
combinationsChoose: /(\d+)\s*(?:choose|C|c)\s*(\d+)/i,
|
|
75
|
+
combinationsFrom:
|
|
76
|
+
/(?:choose|combination|ways\s+to\s+choose)\s+(\d+)\s+(?:from|items?\s+from)\s+(\d+)/i,
|
|
77
|
+
combinationsHowMany: /how\s+many\s+ways\s+(?:to\s+)?choose\s+(\d+)\s+(?:items?\s+)?from\s+(\d+)/i,
|
|
78
|
+
permutationsP: /(\d+)\s*(?:P|p)\s*(\d+)/i,
|
|
79
|
+
permutationsWord: /(?:permutation|arrangement).*?(\d+).*?(\d+)/i,
|
|
80
|
+
lastDigitMod: /(\d+)\s*\^\s*(\d+)\s*mod\s*10|last\s+digit\s+(?:of\s+)?(\d+)\s*\^\s*(\d+)/i,
|
|
81
|
+
} as const;
|
|
82
|
+
|
|
83
|
+
// =============================================================================
|
|
84
|
+
// FORMULA PATTERNS - Tier 4 (Expensive)
|
|
85
|
+
// =============================================================================
|
|
86
|
+
|
|
87
|
+
export const TIER4 = {
|
|
88
|
+
pythagorean: [
|
|
89
|
+
/(?:legs?|sides?)\s*(?:of\s*)?(\d+(?:\.\d+)?)\s*and\s*(\d+(?:\.\d+)?)[^]*?hypoten/i,
|
|
90
|
+
/triangle[^]*?(\d+(?:\.\d+)?)[^]*?(\d+(?:\.\d+)?)[^]*?hypoten/i,
|
|
91
|
+
/right\s*triangle[^]*?(\d+(?:\.\d+)?)[^,]*?(\d+(?:\.\d+)?)/i,
|
|
92
|
+
] as const,
|
|
93
|
+
trailingZeros: /trailing\s+zeros?\s+(?:in\s+)?(\d+)[!]?\s*(?:factorial)?/i,
|
|
94
|
+
geometricSeries: [
|
|
95
|
+
/1\s*\+\s*1\/(\d+)\s*\+\s*1\/\d+\s*\+.*(?:sum|infinite|\.\.\.)/i,
|
|
96
|
+
/sum.*1\s*\+\s*1\/(\d+)\s*\+\s*1\/\d+/i,
|
|
97
|
+
] as const,
|
|
98
|
+
matrixDet:
|
|
99
|
+
/(?:determinant|det).*?\[\s*\[\s*(-?\d+)\s*,\s*(-?\d+)\s*\]\s*,\s*\[\s*(-?\d+)\s*,\s*(-?\d+)\s*\]\s*\]/i,
|
|
100
|
+
compoundInterest:
|
|
101
|
+
/\$?(\d+(?:,\d{3})*(?:\.\d+)?)\s*(?:at|with)\s*(\d+(?:\.\d+)?)\s*%\s*(?:annual\s+)?(?:compound\s+)?interest\s*(?:for\s+)?(\d+)\s*years?/i,
|
|
102
|
+
} as const;
|
|
103
|
+
|
|
104
|
+
// =============================================================================
|
|
105
|
+
// CALCULUS PATTERNS
|
|
106
|
+
// =============================================================================
|
|
107
|
+
|
|
108
|
+
export const CALCULUS = {
|
|
109
|
+
derivative: [
|
|
110
|
+
/(?:derivative|d\/dx)\s+(?:of\s+)?(.+?)\s+(?:at|evaluated\s+at|when)\s+x\s*=\s*(\d+)/i,
|
|
111
|
+
/(?:derivative|d\/dx)\s+(?:of\s+)?([^.?!]+)/i,
|
|
112
|
+
/(?:differentiate)\s+(.+?)\s+(?:at|evaluated\s+at)\s+x\s*=\s*(\d+)/i,
|
|
113
|
+
/(?:differentiate)\s+([^.?!]+)/i,
|
|
114
|
+
] as const,
|
|
115
|
+
integral: [
|
|
116
|
+
/(?:integral|integrate)\s+(?:of\s+)?(.+?)\s+from\s+(-?\d+)\s+to\s+(-?\d+)/i,
|
|
117
|
+
/\u222B\s*(.+?)\s*(?:dx)?\s*from\s*(-?\d+)\s*to\s*(-?\d+)/i,
|
|
118
|
+
] as const,
|
|
119
|
+
polynomial: /([+-]?)(\d*\.?\d*)x(?:\^([+-]?\d+))?|([+-]?)(\d+\.?\d*)/g,
|
|
120
|
+
} as const;
|
|
121
|
+
|
|
122
|
+
// =============================================================================
|
|
123
|
+
// WORD PROBLEM PATTERNS
|
|
124
|
+
// =============================================================================
|
|
125
|
+
|
|
126
|
+
export const WORD_PROBLEM_PATTERNS: WordProblemMatch[] = [
|
|
127
|
+
// Multiplication patterns
|
|
128
|
+
{
|
|
129
|
+
pattern: /twice\s+(?:as\s+(?:many|much)\s+(?:as\s+)?)?(\d+(?:\.\d+)?)/i,
|
|
130
|
+
compute: (m) => (m[1] ? parseFloat(m[1]) * 2 : null),
|
|
131
|
+
method: "word_twice",
|
|
132
|
+
},
|
|
133
|
+
{
|
|
134
|
+
pattern: /(\d+(?:\.\d+)?)\s+times\s+(\d+(?:\.\d+)?)/i,
|
|
135
|
+
compute: (m) => (m[1] && m[2] ? parseFloat(m[1]) * parseFloat(m[2]) : null),
|
|
136
|
+
method: "word_times",
|
|
137
|
+
},
|
|
138
|
+
{
|
|
139
|
+
pattern: /triple\s+(?:of\s+)?(\d+(?:\.\d+)?)/i,
|
|
140
|
+
compute: (m) => (m[1] ? parseFloat(m[1]) * 3 : null),
|
|
141
|
+
method: "word_triple",
|
|
142
|
+
},
|
|
143
|
+
{
|
|
144
|
+
pattern: /double\s+(?:of\s+)?(\d+(?:\.\d+)?)/i,
|
|
145
|
+
compute: (m) => (m[1] ? parseFloat(m[1]) * 2 : null),
|
|
146
|
+
method: "word_double",
|
|
147
|
+
},
|
|
148
|
+
|
|
149
|
+
// Division patterns
|
|
150
|
+
{
|
|
151
|
+
pattern: /half\s+(?:of\s+)?(\d+(?:\.\d+)?)/i,
|
|
152
|
+
compute: (m) => (m[1] ? parseFloat(m[1]) / 2 : null),
|
|
153
|
+
method: "word_half",
|
|
154
|
+
},
|
|
155
|
+
{
|
|
156
|
+
pattern: /(?:one\s+)?third\s+(?:of\s+)?(\d+(?:\.\d+)?)/i,
|
|
157
|
+
compute: (m) => (m[1] ? parseFloat(m[1]) / 3 : null),
|
|
158
|
+
method: "word_third",
|
|
159
|
+
},
|
|
160
|
+
{
|
|
161
|
+
pattern: /(?:one\s+)?quarter\s+(?:of\s+)?(\d+(?:\.\d+)?)/i,
|
|
162
|
+
compute: (m) => (m[1] ? parseFloat(m[1]) / 4 : null),
|
|
163
|
+
method: "word_quarter",
|
|
164
|
+
},
|
|
165
|
+
{
|
|
166
|
+
pattern: /(\d+(?:\.\d+)?)\s+divided\s+by\s+(\d+(?:\.\d+)?)/i,
|
|
167
|
+
compute: (m) => {
|
|
168
|
+
if (!m[1] || !m[2]) return null;
|
|
169
|
+
const b = parseFloat(m[2]);
|
|
170
|
+
return b !== 0 ? parseFloat(m[1]) / b : null;
|
|
171
|
+
},
|
|
172
|
+
method: "word_divide",
|
|
173
|
+
},
|
|
174
|
+
|
|
175
|
+
// Addition patterns
|
|
176
|
+
{
|
|
177
|
+
pattern: /sum\s+of\s+(\d+(?:\.\d+)?)\s+and\s+(\d+(?:\.\d+)?)/i,
|
|
178
|
+
compute: (m) => (m[1] && m[2] ? parseFloat(m[1]) + parseFloat(m[2]) : null),
|
|
179
|
+
method: "word_sum",
|
|
180
|
+
},
|
|
181
|
+
{
|
|
182
|
+
pattern: /(\d+(?:\.\d+)?)\s+plus\s+(\d+(?:\.\d+)?)/i,
|
|
183
|
+
compute: (m) => (m[1] && m[2] ? parseFloat(m[1]) + parseFloat(m[2]) : null),
|
|
184
|
+
method: "word_plus",
|
|
185
|
+
},
|
|
186
|
+
{
|
|
187
|
+
pattern: /(\d+(?:\.\d+)?)\s+added\s+to\s+(\d+(?:\.\d+)?)/i,
|
|
188
|
+
compute: (m) => (m[1] && m[2] ? parseFloat(m[1]) + parseFloat(m[2]) : null),
|
|
189
|
+
method: "word_add",
|
|
190
|
+
},
|
|
191
|
+
{
|
|
192
|
+
pattern: /total\s+of\s+(\d+(?:\.\d+)?)\s+and\s+(\d+(?:\.\d+)?)/i,
|
|
193
|
+
compute: (m) => (m[1] && m[2] ? parseFloat(m[1]) + parseFloat(m[2]) : null),
|
|
194
|
+
method: "word_total",
|
|
195
|
+
},
|
|
196
|
+
|
|
197
|
+
// Subtraction patterns
|
|
198
|
+
{
|
|
199
|
+
pattern: /difference\s+(?:between|of)\s+(\d+(?:\.\d+)?)\s+and\s+(\d+(?:\.\d+)?)/i,
|
|
200
|
+
compute: (m) => (m[1] && m[2] ? Math.abs(parseFloat(m[1]) - parseFloat(m[2])) : null),
|
|
201
|
+
method: "word_difference",
|
|
202
|
+
},
|
|
203
|
+
{
|
|
204
|
+
pattern: /(\d+(?:\.\d+)?)\s+minus\s+(\d+(?:\.\d+)?)/i,
|
|
205
|
+
compute: (m) => (m[1] && m[2] ? parseFloat(m[1]) - parseFloat(m[2]) : null),
|
|
206
|
+
method: "word_minus",
|
|
207
|
+
},
|
|
208
|
+
{
|
|
209
|
+
pattern: /(\d+(?:\.\d+)?)\s+less\s+than\s+(\d+(?:\.\d+)?)/i,
|
|
210
|
+
compute: (m) => (m[1] && m[2] ? parseFloat(m[2]) - parseFloat(m[1]) : null), // Note: reversed!
|
|
211
|
+
method: "word_less_than",
|
|
212
|
+
},
|
|
213
|
+
{
|
|
214
|
+
pattern: /subtract\s+(\d+(?:\.\d+)?)\s+from\s+(\d+(?:\.\d+)?)/i,
|
|
215
|
+
compute: (m) => (m[1] && m[2] ? parseFloat(m[2]) - parseFloat(m[1]) : null), // Note: reversed!
|
|
216
|
+
method: "word_subtract",
|
|
217
|
+
},
|
|
218
|
+
|
|
219
|
+
// Product pattern
|
|
220
|
+
{
|
|
221
|
+
pattern: /product\s+of\s+(\d+(?:\.\d+)?)\s+and\s+(\d+(?:\.\d+)?)/i,
|
|
222
|
+
compute: (m) => (m[1] && m[2] ? parseFloat(m[1]) * parseFloat(m[2]) : null),
|
|
223
|
+
method: "word_product",
|
|
224
|
+
},
|
|
225
|
+
|
|
226
|
+
// Quotient pattern
|
|
227
|
+
{
|
|
228
|
+
pattern: /quotient\s+of\s+(\d+(?:\.\d+)?)\s+and\s+(\d+(?:\.\d+)?)/i,
|
|
229
|
+
compute: (m) => {
|
|
230
|
+
if (!m[1] || !m[2]) return null;
|
|
231
|
+
const b = parseFloat(m[2]);
|
|
232
|
+
return b !== 0 ? parseFloat(m[1]) / b : null;
|
|
233
|
+
},
|
|
234
|
+
method: "word_quotient",
|
|
235
|
+
},
|
|
236
|
+
|
|
237
|
+
// "X more than Y"
|
|
238
|
+
{
|
|
239
|
+
pattern: /(\d+(?:\.\d+)?)\s+more\s+than\s+(\d+(?:\.\d+)?)/i,
|
|
240
|
+
compute: (m) => (m[1] && m[2] ? parseFloat(m[2]) + parseFloat(m[1]) : null),
|
|
241
|
+
method: "word_more_than",
|
|
242
|
+
},
|
|
243
|
+
|
|
244
|
+
// Squared / Cubed
|
|
245
|
+
{
|
|
246
|
+
pattern: /(\d+(?:\.\d+)?)\s+squared/i,
|
|
247
|
+
compute: (m) => (m[1] ? parseFloat(m[1]) ** 2 : null),
|
|
248
|
+
method: "word_squared",
|
|
249
|
+
},
|
|
250
|
+
{
|
|
251
|
+
pattern: /(\d+(?:\.\d+)?)\s+cubed/i,
|
|
252
|
+
compute: (m) => (m[1] ? parseFloat(m[1]) ** 3 : null),
|
|
253
|
+
method: "word_cubed",
|
|
254
|
+
},
|
|
255
|
+
|
|
256
|
+
// Rate calculations
|
|
257
|
+
{
|
|
258
|
+
pattern:
|
|
259
|
+
/(\d+(?:\.\d+)?)\s+(?:items?|things?|units?)?\s*(?:at|for)\s+\$?(\d+(?:\.\d+)?)\s+(?:each|per|apiece)/i,
|
|
260
|
+
compute: (m) => (m[1] && m[2] ? parseFloat(m[1]) * parseFloat(m[2]) : null),
|
|
261
|
+
method: "word_rate",
|
|
262
|
+
},
|
|
263
|
+
|
|
264
|
+
// Average of numbers
|
|
265
|
+
{
|
|
266
|
+
pattern: /average\s+of\s+([\d,\s]+(?:and\s+\d+)?)/i,
|
|
267
|
+
compute: (m) => {
|
|
268
|
+
if (!m[1]) return null;
|
|
269
|
+
const nums = m[1].match(/\d+(?:\.\d+)?/g);
|
|
270
|
+
if (!nums || nums.length === 0) return null;
|
|
271
|
+
const sum = nums.reduce((a, b) => a + parseFloat(b), 0);
|
|
272
|
+
return sum / nums.length;
|
|
273
|
+
},
|
|
274
|
+
method: "word_average",
|
|
275
|
+
},
|
|
276
|
+
];
|
|
277
|
+
|
|
278
|
+
// =============================================================================
|
|
279
|
+
// MULTI-STEP WORD PROBLEM PATTERNS
|
|
280
|
+
// =============================================================================
|
|
281
|
+
|
|
282
|
+
export const MULTI_STEP = {
|
|
283
|
+
twice: /(\b[A-Z][a-z]+\b)\s+has\s+twice\s+(?:as\s+many\s+(?:as\s+)?)?(\b[A-Z][a-z]+\b)/gi,
|
|
284
|
+
half: /(\b[A-Z][a-z]+\b)\s+has\s+half\s+(?:as\s+many\s+(?:as\s+)?)?(\b[A-Z][a-z]+\b)/gi,
|
|
285
|
+
more: /(\b[A-Z][a-z]+\b)\s+has\s+(\d+)\s+more\s+than\s+(\b[A-Z][a-z]+\b)/gi,
|
|
286
|
+
less: /(\b[A-Z][a-z]+\b)\s+has\s+(\d+)\s+(?:less|fewer)\s+than\s+(\b[A-Z][a-z]+\b)/gi,
|
|
287
|
+
triple:
|
|
288
|
+
/(\b[A-Z][a-z]+\b)\s+has\s+(?:three|triple)\s+(?:times\s+)?(?:as\s+many\s+(?:as\s+)?)?(\b[A-Z][a-z]+\b)/gi,
|
|
289
|
+
directValue: /(\b[A-Z][a-z]+\b)\s+(?:has|have|had|owns?|bought|got|earned)\s+(\d+(?:\.\d+)?)/gi,
|
|
290
|
+
question: /(?:how\s+many|what)\s+(?:does|do)\s+(\b[A-Z][a-z]+\b)\s+have/i,
|
|
291
|
+
} as const;
|
|
292
|
+
|
|
293
|
+
// =============================================================================
|
|
294
|
+
// EXTRACT PATTERNS
|
|
295
|
+
// =============================================================================
|
|
296
|
+
|
|
297
|
+
export const EXTRACT = {
|
|
298
|
+
binaryOp: /\b(\d+(?:\.\d+)?)\s*([+\-*/])\s*(\d+(?:\.\d+)?)\b/g,
|
|
299
|
+
} as const;
|
|
300
|
+
|
|
301
|
+
// =============================================================================
|
|
302
|
+
// CONFIDENCE PATTERNS
|
|
303
|
+
// =============================================================================
|
|
304
|
+
|
|
305
|
+
export const CONFIDENCE_POSITIVE: WeightedSignal[] = [
|
|
306
|
+
// Very high confidence (0.9+)
|
|
307
|
+
{ pattern: /^[\d\s+\-*/().]+$/, weight: 0.95, name: "pure_arithmetic" },
|
|
308
|
+
{ pattern: /what\s+is\s+\d+\s*[+\-*/]\s*\d+/i, weight: 0.95, name: "explicit_arithmetic" },
|
|
309
|
+
{ pattern: /calculate\s+\d+/i, weight: 0.9, name: "calculate_number" },
|
|
310
|
+
{ pattern: /\d+!/i, weight: 0.9, name: "factorial" },
|
|
311
|
+
{ pattern: /fibonacci\s*\d+|(\d+)(?:st|nd|rd|th)\s*fibonacci/i, weight: 0.9, name: "fibonacci" },
|
|
312
|
+
|
|
313
|
+
// High confidence (0.7-0.9)
|
|
314
|
+
{ pattern: /sqrt|square\s*root|\u221A/i, weight: 0.85, name: "sqrt" },
|
|
315
|
+
{ pattern: /log[\u2081\u2080]?\s*\(?\d/i, weight: 0.85, name: "logarithm" },
|
|
316
|
+
{ pattern: /\d+\s*\^\s*\d+|\d+\s*\*\*\s*\d+/i, weight: 0.85, name: "power" },
|
|
317
|
+
{ pattern: /\d+\s*%\s*of\s*\d+/i, weight: 0.85, name: "percentage" },
|
|
318
|
+
{ pattern: /gcd|lcm/i, weight: 0.85, name: "gcd_lcm" },
|
|
319
|
+
{ pattern: /\d+\s*mod(ulo)?\s*\d+/i, weight: 0.85, name: "modulo" },
|
|
320
|
+
{ pattern: /is\s+\d+\s+(?:a\s+)?prime/i, weight: 0.85, name: "primality" },
|
|
321
|
+
{ pattern: /legs?\s+\d+\s+and\s+\d+.*hypoten/i, weight: 0.85, name: "pythagorean" },
|
|
322
|
+
{ pattern: /x[\u00B22]\s*[+-]\s*\d+x\s*[+-]\s*\d+\s*=\s*0/i, weight: 0.85, name: "quadratic" },
|
|
323
|
+
{ pattern: /trailing\s+zeros?\s+.*\d+[!]/i, weight: 0.85, name: "trailing_zeros" },
|
|
324
|
+
{ pattern: /last\s+digit.*\^\d+|\d+\s*\^\s*\d+\s*mod\s*10/i, weight: 0.85, name: "last_digit" },
|
|
325
|
+
// Calculus
|
|
326
|
+
{ pattern: /derivative.*x\^?\d|d\/dx/i, weight: 0.85, name: "derivative" },
|
|
327
|
+
{ pattern: /integral.*from.*to|integrate.*from/i, weight: 0.85, name: "definite_integral" },
|
|
328
|
+
// Combinatorics
|
|
329
|
+
{ pattern: /\d+\s*choose\s*\d+/i, weight: 0.9, name: "combinations" },
|
|
330
|
+
{ pattern: /\d+\s*[CP]\s*\d+/i, weight: 0.85, name: "perm_comb" },
|
|
331
|
+
// Matrix
|
|
332
|
+
{ pattern: /determinant.*\[\[/i, weight: 0.85, name: "determinant" },
|
|
333
|
+
// Compound interest
|
|
334
|
+
{ pattern: /compound\s+interest.*\d+\s*%.*\d+\s*year/i, weight: 0.85, name: "compound_interest" },
|
|
335
|
+
// Math facts (rationality)
|
|
336
|
+
{ pattern: /rational\s+or\s+irrational/i, weight: 0.95, name: "rationality" },
|
|
337
|
+
{
|
|
338
|
+
pattern: /is\s+(?:sqrt|√|pi|e|phi)\b.*\b(?:rational|irrational)/i,
|
|
339
|
+
weight: 0.95,
|
|
340
|
+
name: "known_rationality",
|
|
341
|
+
},
|
|
342
|
+
|
|
343
|
+
// Logic patterns (modus ponens, modus tollens, syllogism, XOR)
|
|
344
|
+
{
|
|
345
|
+
pattern: /if\s+.+?,?\s+(?:then\s+)?.+?\.\s+.+?\.\s+(?:is\s+)?/i,
|
|
346
|
+
weight: 0.9,
|
|
347
|
+
name: "conditional_logic",
|
|
348
|
+
},
|
|
349
|
+
{
|
|
350
|
+
pattern: /all\s+\w+\s+are\s+\w+.*all\s+\w+\s+are\s+\w+.*valid/i,
|
|
351
|
+
weight: 0.95,
|
|
352
|
+
name: "syllogism",
|
|
353
|
+
},
|
|
354
|
+
{ pattern: /exclusive.*both|both.*exclusive/i, weight: 0.9, name: "xor_violation" },
|
|
355
|
+
{ pattern: /yes\s+or\s+no.*\?/i, weight: 0.8, name: "yes_no_question" },
|
|
356
|
+
|
|
357
|
+
// Probability patterns (independent events, gambler's fallacy)
|
|
358
|
+
{
|
|
359
|
+
pattern: /fair\s+coin.*(?:probability|chance)/i,
|
|
360
|
+
weight: 0.95,
|
|
361
|
+
name: "fair_coin_prob",
|
|
362
|
+
},
|
|
363
|
+
{
|
|
364
|
+
pattern: /independent.*(?:probability|chance)/i,
|
|
365
|
+
weight: 0.9,
|
|
366
|
+
name: "independent_event",
|
|
367
|
+
},
|
|
368
|
+
{
|
|
369
|
+
pattern: /in\s+a\s+row.*(?:probability|chance|what['']?s)/i,
|
|
370
|
+
weight: 0.85,
|
|
371
|
+
name: "streak_probability",
|
|
372
|
+
},
|
|
373
|
+
{
|
|
374
|
+
pattern: /(?:heads|tails)\s*\d+\s*(?:times?)?.*(?:next|probability)/i,
|
|
375
|
+
weight: 0.9,
|
|
376
|
+
name: "coin_streak",
|
|
377
|
+
},
|
|
378
|
+
|
|
379
|
+
// Medium confidence (0.5-0.7)
|
|
380
|
+
{ pattern: /twice\s+\d+/i, weight: 0.75, name: "twice_number" },
|
|
381
|
+
{ pattern: /half\s+(?:of\s+)?\d+/i, weight: 0.75, name: "half_number" },
|
|
382
|
+
{ pattern: /\d+\s+(?:plus|minus|times)\s+\d+/i, weight: 0.75, name: "word_operation" },
|
|
383
|
+
{ pattern: /sum\s+of\s+\d+\s+and\s+\d+/i, weight: 0.75, name: "sum_of" },
|
|
384
|
+
{ pattern: /average\s+of.*\d+/i, weight: 0.7, name: "average" },
|
|
385
|
+
{ pattern: /\d+\s+(?:more|less)\s+than\s+\d+/i, weight: 0.7, name: "more_less" },
|
|
386
|
+
|
|
387
|
+
// Lower confidence (multi-step word problems)
|
|
388
|
+
{ pattern: /[A-Z][a-z]+\s+has\s+twice.*[A-Z][a-z]+/i, weight: 0.6, name: "entity_twice" },
|
|
389
|
+
{ pattern: /[A-Z][a-z]+\s+has\s+\d+\s+more\s+than/i, weight: 0.6, name: "entity_more" },
|
|
390
|
+
{ pattern: /how\s+many\s+does\s+[A-Z][a-z]+\s+have/i, weight: 0.5, name: "entity_question" },
|
|
391
|
+
{ pattern: /total|altogether|combined/i, weight: 0.5, name: "total_question" },
|
|
392
|
+
|
|
393
|
+
// Generic signals (low weight)
|
|
394
|
+
{ pattern: /calculate/i, weight: 0.4, name: "calculate" },
|
|
395
|
+
{ pattern: /compute/i, weight: 0.4, name: "compute" },
|
|
396
|
+
{ pattern: /what\s+is/i, weight: 0.3, name: "what_is" },
|
|
397
|
+
{ pattern: /\d+\s*[+\-*/]\s*\d+/, weight: 0.5, name: "has_operation" },
|
|
398
|
+
];
|
|
399
|
+
|
|
400
|
+
export const CONFIDENCE_NEGATIVE: NegativeSignal[] = [
|
|
401
|
+
{ pattern: /prove/i, penalty: 0.8, name: "prove" },
|
|
402
|
+
{ pattern: /why/i, penalty: 0.7, name: "why" },
|
|
403
|
+
{ pattern: /explain/i, penalty: 0.7, name: "explain" },
|
|
404
|
+
{ pattern: /derive/i, penalty: 0.6, name: "derive" },
|
|
405
|
+
{ pattern: /show\s+(that|how|why)/i, penalty: 0.6, name: "show" },
|
|
406
|
+
{ pattern: /compare/i, penalty: 0.5, name: "compare" },
|
|
407
|
+
{ pattern: /what\s+is\s+the\s+best/i, penalty: 0.5, name: "best" },
|
|
408
|
+
{ pattern: /should/i, penalty: 0.4, name: "should" },
|
|
409
|
+
// Removed: rational/irrational penalty - now handled by math facts solver
|
|
410
|
+
{ pattern: /true\s+or\s+false/i, penalty: 0.8, name: "boolean" },
|
|
411
|
+
{ pattern: /infinite.*(?!sum|series)/i, penalty: 0.5, name: "infinite_general" },
|
|
412
|
+
];
|
|
413
|
+
|
|
414
|
+
// =============================================================================
|
|
415
|
+
// LIKELY COMPUTABLE PATTERNS
|
|
416
|
+
// =============================================================================
|
|
417
|
+
|
|
418
|
+
export const LIKELY_COMPUTABLE_POSITIVE = [
|
|
419
|
+
/what\s+is\s+\d/i,
|
|
420
|
+
/what\s+is\s+x[\u00B22^]/i,
|
|
421
|
+
/calculate/i,
|
|
422
|
+
/compute/i,
|
|
423
|
+
/\d+\s*[+\-*/]\s*\d+/,
|
|
424
|
+
/fibonacci/i,
|
|
425
|
+
/factorial|\d+!/,
|
|
426
|
+
/sqrt|square\s*root|\u221A/i,
|
|
427
|
+
/log[\u2081\u2080]?\s*\(?/i,
|
|
428
|
+
/\^|to\s+the\s+power/i,
|
|
429
|
+
/%\s*of\s*\d/i,
|
|
430
|
+
/hypoten/i,
|
|
431
|
+
/quadratic/i,
|
|
432
|
+
/gcd|lcm/i,
|
|
433
|
+
/mod(ulo)?/i,
|
|
434
|
+
/=\s*0.*root/i,
|
|
435
|
+
/larger\s+root|smaller\s+root/i,
|
|
436
|
+
/legs?\s+\d+\s+and\s+\d+/i,
|
|
437
|
+
/is\s+\d+\s+(?:a\s+)?prime/i,
|
|
438
|
+
/twice\s+(?:as\s+)?/i,
|
|
439
|
+
/half\s+(?:of\s+)?/i,
|
|
440
|
+
/double\s+(?:of\s+)?/i,
|
|
441
|
+
/triple\s+(?:of\s+)?/i,
|
|
442
|
+
/sum\s+of\s+\d+\s+and\s+\d+/i,
|
|
443
|
+
/difference\s+(?:between|of)/i,
|
|
444
|
+
/product\s+of\s+\d+\s+and\s+\d+/i,
|
|
445
|
+
/quotient\s+of/i,
|
|
446
|
+
/\d+\s+(?:plus|minus|times)\s+\d+/i,
|
|
447
|
+
/\d+\s+(?:divided\s+by|multiplied\s+by)\s+\d+/i,
|
|
448
|
+
/\d+\s+(?:more|less)\s+than\s+\d+/i,
|
|
449
|
+
/\d+\s+squared|\d+\s+cubed/i,
|
|
450
|
+
/average\s+of/i,
|
|
451
|
+
/third\s+(?:of\s+)?/i,
|
|
452
|
+
/quarter\s+(?:of\s+)?/i,
|
|
453
|
+
/trailing\s+zeros?/i,
|
|
454
|
+
/infinite\s+(?:series|sum)|geometric\s+series/i,
|
|
455
|
+
/last\s+digit/i,
|
|
456
|
+
/derivative|d\/dx|differentiate/i,
|
|
457
|
+
/integral|integrate|\u222B/i,
|
|
458
|
+
/\d+\s*choose\s*\d+/i,
|
|
459
|
+
/combination|permutation/i,
|
|
460
|
+
/ways\s+to\s+choose/i,
|
|
461
|
+
/determinant|det\s*\(/i,
|
|
462
|
+
/compound\s+interest/i,
|
|
463
|
+
// Math facts (rationality questions)
|
|
464
|
+
/rational\s+or\s+irrational/i,
|
|
465
|
+
/is\s+(?:sqrt|√|pi|e|phi)\b.*\b(?:rational|irrational)/i,
|
|
466
|
+
// Logic patterns (modus ponens, modus tollens, syllogism, XOR)
|
|
467
|
+
/if\s+[^,]+,\s*[^.]+\.\s*[^.]+\.\s*is\s+/i, // If X, Y. Z. Is ...?
|
|
468
|
+
/all\s+\w+\s+are\s+\w+.*valid/i,
|
|
469
|
+
/exclusive.*both.*violated/i,
|
|
470
|
+
// Probability patterns (independent events, gambler's fallacy, birthday paradox)
|
|
471
|
+
/fair\s+coin.*(?:probability|chance)/i,
|
|
472
|
+
/independent.*(?:probability|chance)/i,
|
|
473
|
+
/in\s+a\s+row.*(?:probability|chance|what['']?s)/i,
|
|
474
|
+
/birthday.*share|share.*birthday/i,
|
|
475
|
+
/\d+\s*(?:people|persons?).*birthday/i,
|
|
476
|
+
] as const;
|
|
477
|
+
|
|
478
|
+
export const LIKELY_COMPUTABLE_NEGATIVE = [
|
|
479
|
+
/prove/i,
|
|
480
|
+
/why/i,
|
|
481
|
+
/explain/i,
|
|
482
|
+
/derive/i,
|
|
483
|
+
/show\s+(that|how|why)/i,
|
|
484
|
+
/what\s+is\s+the\s+best/i,
|
|
485
|
+
/should/i,
|
|
486
|
+
/compare/i,
|
|
487
|
+
// Removed: /rational\s+or\s+irrational/i - now handled by math facts solver
|
|
488
|
+
/true\s+or\s+false/i,
|
|
489
|
+
] as const;
|
|
490
|
+
|
|
491
|
+
// =============================================================================
|
|
492
|
+
// SPAN PATTERNS - For multi-expression extraction (combined regex approach)
|
|
493
|
+
// These patterns can identify computable spans within arbitrary text.
|
|
494
|
+
// Used by extractAndCompute() for O(n) text scanning.
|
|
495
|
+
// =============================================================================
|
|
496
|
+
|
|
497
|
+
export interface SpanPattern {
|
|
498
|
+
/** Regex to match the span (should capture the full computable expression) */
|
|
499
|
+
regex: RegExp;
|
|
500
|
+
/** Name for debugging/tracking */
|
|
501
|
+
name: string;
|
|
502
|
+
}
|
|
503
|
+
|
|
504
|
+
/**
|
|
505
|
+
* Patterns suitable for span extraction from text.
|
|
506
|
+
* Each pattern should match a complete, self-contained computable expression.
|
|
507
|
+
* Excludes patterns that require question context (e.g., "is X prime?")
|
|
508
|
+
*/
|
|
509
|
+
export const SPAN_PATTERNS: SpanPattern[] = [
|
|
510
|
+
// Tier 1: Ultra-fast
|
|
511
|
+
{ regex: /(\d+(?:\.\d+)?)\s*%\s*of\s*(\d+(?:\.\d+)?)/i, name: "percentage" },
|
|
512
|
+
{ regex: /(\d+)!/i, name: "factorial_symbol" },
|
|
513
|
+
{ regex: /factorial\s*(?:of\s*)?(\d+)/i, name: "factorial_word" },
|
|
514
|
+
|
|
515
|
+
// Tier 2: Fast
|
|
516
|
+
{ regex: /sqrt\s*\(\s*(\d+(?:\.\d+)?)\s*\)/i, name: "sqrt_parens" },
|
|
517
|
+
{ regex: /√(\d+(?:\.\d+)?)/i, name: "sqrt_symbol" },
|
|
518
|
+
{ regex: /square\s+root\s+of\s+(\d+(?:\.\d+)?)/i, name: "sqrt_words" },
|
|
519
|
+
{ regex: /(\d+(?:\.\d+)?)\s*\^\s*(\d+(?:\.\d+)?)/i, name: "power_caret" },
|
|
520
|
+
{ regex: /(\d+(?:\.\d+)?)\s*\*\*\s*(\d+(?:\.\d+)?)/i, name: "power_stars" },
|
|
521
|
+
{ regex: /gcd\s*\(\s*(\d+)\s*,\s*(\d+)\s*\)/i, name: "gcd" },
|
|
522
|
+
{ regex: /lcm\s*\(\s*(\d+)\s*,\s*(\d+)\s*\)/i, name: "lcm" },
|
|
523
|
+
|
|
524
|
+
// Tier 3: Medium
|
|
525
|
+
{ regex: /(\d+)\s*choose\s*(\d+)/i, name: "combinations" },
|
|
526
|
+
{ regex: /(\d+)\s*C\s*(\d+)/i, name: "combinations_c" },
|
|
527
|
+
{ regex: /(\d+)\s*P\s*(\d+)/i, name: "permutations" },
|
|
528
|
+
{ regex: /log\s*\(\s*(\d+(?:\.\d+)?)\s*\)/i, name: "log" },
|
|
529
|
+
{ regex: /ln\s*\(\s*(\d+(?:\.\d+)?)\s*\)/i, name: "ln" },
|
|
530
|
+
|
|
531
|
+
// Tier 4: Expensive
|
|
532
|
+
{ regex: /det\s*\(\s*\[\[[\d,\s\-[\]]+\]\]\s*\)/i, name: "determinant" },
|
|
533
|
+
{ regex: /determinant\s+of\s+\[\[[\d,\s\-[\]]+\]\]/i, name: "determinant_of" },
|
|
534
|
+
|
|
535
|
+
// Calculus
|
|
536
|
+
{
|
|
537
|
+
regex: /derivative\s+of\s+([x\d\s+\-*^]+)\s+at\s+x\s*=\s*(\d+)/i,
|
|
538
|
+
name: "derivative_at",
|
|
539
|
+
},
|
|
540
|
+
{
|
|
541
|
+
regex: /d\/dx\s*\(\s*([x\d\s+\-*^]+)\s*\)\s*at\s+x\s*=\s*(\d+)/i,
|
|
542
|
+
name: "derivative_dx_at",
|
|
543
|
+
},
|
|
544
|
+
{
|
|
545
|
+
regex: /integrate\s+([x\d\s+\-*^]+)\s+from\s+(-?\d+)\s+to\s+(-?\d+)/i,
|
|
546
|
+
name: "integral_from_to",
|
|
547
|
+
},
|
|
548
|
+
{
|
|
549
|
+
regex: /integral\s+of\s+([x\d\s+\-*^]+)\s+from\s+(-?\d+)\s+to\s+(-?\d+)/i,
|
|
550
|
+
name: "integral_of_from_to",
|
|
551
|
+
},
|
|
552
|
+
];
|
|
553
|
+
|
|
554
|
+
/**
|
|
555
|
+
* Build combined regex for O(n) scanning.
|
|
556
|
+
* Uses alternation which V8 optimizes into an efficient automaton.
|
|
557
|
+
*/
|
|
558
|
+
export function buildCombinedSpanRegex(): RegExp {
|
|
559
|
+
const combined = SPAN_PATTERNS.map((p) => `(?:${p.regex.source})`).join("|");
|
|
560
|
+
return new RegExp(combined, "gi");
|
|
561
|
+
}
|
|
562
|
+
|
|
563
|
+
/** Pre-built combined regex for span extraction */
|
|
564
|
+
export const COMBINED_SPAN_REGEX = buildCombinedSpanRegex();
|