@hviana/sema 0.5.2 → 0.5.3
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/AGENTS.md +114 -52
- package/HOW_IT_WORKS.md +275 -184
- package/dist/src/mind/bridge.d.ts +5 -7
- package/dist/src/mind/bridge.js +6 -97
- package/dist/src/mind/match.d.ts +159 -0
- package/dist/src/mind/match.js +300 -7
- package/dist/src/mind/mechanisms/prefix-completion.d.ts +22 -0
- package/dist/src/mind/{prefix-completion.js → mechanisms/prefix-completion.js} +64 -91
- package/dist/src/mind/mechanisms/recall.js +10 -108
- package/dist/src/mind/mechanisms/reference.d.ts +6 -0
- package/dist/src/mind/mechanisms/reference.js +296 -0
- package/dist/src/mind/mind.d.ts +1 -1
- package/dist/src/mind/pipeline-mechanism.d.ts +56 -1
- package/dist/src/mind/pipeline-mechanism.js +104 -3
- package/dist/src/mind/pipeline.d.ts +1 -1
- package/dist/src/mind/pipeline.js +13 -1
- package/dist/src/mind/traverse.d.ts +38 -0
- package/dist/src/mind/traverse.js +91 -1
- package/dist/src/store.d.ts +4 -4
- package/jsr.json +6 -0
- package/package.json +1 -1
- package/src/mind/bridge.ts +10 -104
- package/src/mind/match.ts +416 -7
- package/src/mind/{prefix-completion.ts → mechanisms/prefix-completion.ts} +66 -92
- package/src/mind/mechanisms/recall.ts +9 -126
- package/src/mind/mechanisms/reference.ts +343 -0
- package/src/mind/mind.ts +12 -8
- package/src/mind/pipeline-mechanism.ts +120 -3
- package/src/mind/pipeline.ts +16 -2
- package/src/mind/traverse.ts +92 -1
- package/src/store.ts +13 -4
- package/test/33-multi-candidate.test.mjs +21 -11
- package/test/70-prefix-completion.test.mjs +1 -1
- package/test/72-prefix-candidate-supply.test.mjs +7 -9
- package/test/74-prefix-trap-not-sprung-early.test.mjs +1 -1
- package/test/76-reference-binding.test.mjs +471 -0
- package/dist/src/mind/frame-filler.d.ts +0 -15
- package/dist/src/mind/frame-filler.js +0 -535
- package/dist/src/mind/prefix-completion.d.ts +0 -59
- package/src/mind/frame-filler.ts +0 -604
- package/test/69-frame-filler.test.mjs +0 -115
package/dist/src/mind/bridge.js
CHANGED
|
@@ -97,7 +97,7 @@ import { chainReach, leafIdRun } from "./canonical.js";
|
|
|
97
97
|
import { allWindowsAreScaffolding, corpusN, edgeAncestors, hubBound, sharedReachMemo, } from "./traverse.js";
|
|
98
98
|
import { rItem, rNode } from "./trace.js";
|
|
99
99
|
import { junctionContainersFrom } from "./junction.js";
|
|
100
|
-
import { spanHalo } from "./match.js";
|
|
100
|
+
import { alignAround, spanHalo } from "./match.js";
|
|
101
101
|
/** True when some query byte-range left UNACCOUNTED by `spans` contains a
|
|
102
102
|
* STORED window — content the store has seen that the proposed reading
|
|
103
103
|
* simply ignores. The IGNORED-KNOWN principle: a span may be dismissed
|
|
@@ -121,102 +121,11 @@ export function dismissedKnownContent(ctx, query, spans) {
|
|
|
121
121
|
}
|
|
122
122
|
return false;
|
|
123
123
|
}
|
|
124
|
-
|
|
125
|
-
|
|
126
|
-
|
|
127
|
-
|
|
128
|
-
|
|
129
|
-
function align(ctx, q, c, qo, co) {
|
|
130
|
-
const W = ctx.space.maxGroup;
|
|
131
|
-
const reachCap = chainReach(W);
|
|
132
|
-
// Maximal run around the seed.
|
|
133
|
-
let qs = qo, ss = co;
|
|
134
|
-
while (qs > 0 && ss > 0 && q[qs - 1] === c[ss - 1]) {
|
|
135
|
-
qs--;
|
|
136
|
-
ss--;
|
|
137
|
-
}
|
|
138
|
-
let qe = qo, se = co;
|
|
139
|
-
while (qe < q.length && se < c.length && q[qe] === c[se]) {
|
|
140
|
-
qe++;
|
|
141
|
-
se++;
|
|
142
|
-
}
|
|
143
|
-
const matched = [[qs, qe]];
|
|
144
|
-
const gaps = [];
|
|
145
|
-
// The next common run of ≥ W bytes past (qi, si), with each side's gap
|
|
146
|
-
// bounded by chainReach; smallest total gap wins (nearest continuation).
|
|
147
|
-
const runLenAt = (qi, si) => {
|
|
148
|
-
let n = 0;
|
|
149
|
-
while (qi + n < q.length && si + n < c.length && q[qi + n] === c[si + n]) {
|
|
150
|
-
n++;
|
|
151
|
-
}
|
|
152
|
-
return n;
|
|
153
|
-
};
|
|
154
|
-
// RIGHT sweep.
|
|
155
|
-
let qi = qe, si = se;
|
|
156
|
-
for (;;) {
|
|
157
|
-
let found = false;
|
|
158
|
-
for (let total = 1; total <= 2 * reachCap && !found; total++) {
|
|
159
|
-
for (let gq = 0; gq <= Math.min(total, reachCap); gq++) {
|
|
160
|
-
const gs = total - gq;
|
|
161
|
-
if (gs > reachCap)
|
|
162
|
-
continue;
|
|
163
|
-
if (qi + gq >= q.length || si + gs >= c.length)
|
|
164
|
-
continue;
|
|
165
|
-
const n = runLenAt(qi + gq, si + gs);
|
|
166
|
-
if (n >= W || qi + gq + n === q.length) {
|
|
167
|
-
if (n === 0)
|
|
168
|
-
continue;
|
|
169
|
-
if (gq > 0 || gs > 0) {
|
|
170
|
-
gaps.push({ qs: qi, qe: qi + gq, cs: si, ce: si + gs });
|
|
171
|
-
}
|
|
172
|
-
matched.push([qi + gq, qi + gq + n]);
|
|
173
|
-
qi = qi + gq + n;
|
|
174
|
-
si = si + gs + n;
|
|
175
|
-
found = true;
|
|
176
|
-
break;
|
|
177
|
-
}
|
|
178
|
-
}
|
|
179
|
-
}
|
|
180
|
-
if (!found)
|
|
181
|
-
break;
|
|
182
|
-
}
|
|
183
|
-
// LEFT sweep (mirror).
|
|
184
|
-
qi = qs;
|
|
185
|
-
si = ss;
|
|
186
|
-
for (;;) {
|
|
187
|
-
let found = false;
|
|
188
|
-
for (let total = 1; total <= 2 * reachCap && !found; total++) {
|
|
189
|
-
for (let gq = 0; gq <= Math.min(total, reachCap); gq++) {
|
|
190
|
-
const gs = total - gq;
|
|
191
|
-
if (gs > reachCap)
|
|
192
|
-
continue;
|
|
193
|
-
if (qi - gq <= 0 || si - gs <= 0)
|
|
194
|
-
continue;
|
|
195
|
-
// Run ENDING at (qi - gq, si - gs).
|
|
196
|
-
let n = 0;
|
|
197
|
-
while (n < qi - gq && n < si - gs &&
|
|
198
|
-
q[qi - gq - 1 - n] === c[si - gs - 1 - n]) {
|
|
199
|
-
n++;
|
|
200
|
-
}
|
|
201
|
-
if (n >= W || n === qi - gq) {
|
|
202
|
-
if (n === 0)
|
|
203
|
-
continue;
|
|
204
|
-
if (gq > 0 || gs > 0) {
|
|
205
|
-
gaps.push({ qs: qi - gq, qe: qi, cs: si - gs, ce: si });
|
|
206
|
-
}
|
|
207
|
-
matched.push([qi - gq - n, qi - gq]);
|
|
208
|
-
qi = qi - gq - n;
|
|
209
|
-
si = si - gs - n;
|
|
210
|
-
found = true;
|
|
211
|
-
break;
|
|
212
|
-
}
|
|
213
|
-
}
|
|
214
|
-
}
|
|
215
|
-
if (!found)
|
|
216
|
-
break;
|
|
217
|
-
}
|
|
218
|
-
return { matched, gaps };
|
|
219
|
-
}
|
|
124
|
+
// The seeded aligner this file used to own now lives in the shared match
|
|
125
|
+
// family as {@link alignAround} — the frame reading (match.ts) reads the same
|
|
126
|
+
// gaps and asks the OPPOSITE question of them (see AlignGap's own doc). Two
|
|
127
|
+
// consumers, one definition (AGENTS §2.5); the bridge's reading is unchanged.
|
|
128
|
+
const align = alignAround;
|
|
220
129
|
/** Recall's corroborated-substitution bridge — see the module comment.
|
|
221
130
|
* Returns the best bridged grounding proposal, or null. */
|
|
222
131
|
/** `proposed` is a THUNK, not a list: the bridge's own cheap gates (the
|
package/dist/src/mind/match.d.ts
CHANGED
|
@@ -57,6 +57,165 @@ export interface GradedRun {
|
|
|
57
57
|
* (optional — when absent, only literal alignment fires and graded degrades
|
|
58
58
|
* to the original behaviour). Context sites are recognised internally. */
|
|
59
59
|
export declare function alignGraded(ctx: MindContext, query: Uint8Array, contextBytes: Uint8Array, querySites?: ReadonlyArray<Site>): GradedRun[];
|
|
60
|
+
/** One place two byte streams DISAGREE, between runs where they agree: the
|
|
61
|
+
* query span `[qs,qe)` standing where the candidate's `[cs,ce)` stands.
|
|
62
|
+
*
|
|
63
|
+
* Two mechanisms read the same gap and ask OPPOSITE questions of it, which is
|
|
64
|
+
* why the shape lives here rather than in either of them:
|
|
65
|
+
*
|
|
66
|
+
* • the substitution bridge asks whether the two sides MEAN THE SAME, and
|
|
67
|
+
* so EXPANDS the gap (absorbing flanking matched bytes) until the query
|
|
68
|
+
* side is corpus-attested and the pair clears the concept bar;
|
|
69
|
+
* • the frame reading asks WHERE THE SLOT IS, and so CONTRACTS it
|
|
70
|
+
* ({@link contractGap}) until the two sides share nothing at all.
|
|
71
|
+
*
|
|
72
|
+
* Neither reading is derivable from the other, and both need the same gap. */
|
|
73
|
+
export interface AlignGap {
|
|
74
|
+
qs: number;
|
|
75
|
+
qe: number;
|
|
76
|
+
cs: number;
|
|
77
|
+
ce: number;
|
|
78
|
+
}
|
|
79
|
+
/** Extend a seed match (query offset qo ↔ candidate offset co) to its maximal
|
|
80
|
+
* common run, then walk outward in both directions collecting further common
|
|
81
|
+
* runs of at least W bytes across bounded mismatch gaps (each side ≤
|
|
82
|
+
* chainReach). Returns the matched query spans and the mismatch pairs
|
|
83
|
+
* between consecutive runs.
|
|
84
|
+
*
|
|
85
|
+
* This is the SEEDED aligner, distinct from {@link alignRuns}: that one finds
|
|
86
|
+
* every run two structures share anywhere (a weave), this one reads two
|
|
87
|
+
* streams as ONE structure that diverges in bounded places (a frame with
|
|
88
|
+
* slots).
|
|
89
|
+
*
|
|
90
|
+
* Gaps come back in SWEEP order (right sweep, then left), not query order,
|
|
91
|
+
* and only the INTERIOR ones are reported — a consumer that needs the query's
|
|
92
|
+
* unmatched head or tail derives it from `matched`. Both are the bridge's
|
|
93
|
+
* contract, which prices its edges separately (see its matchStart/matchEnd
|
|
94
|
+
* window test); {@link frameSlots} takes the other reading. */
|
|
95
|
+
export declare function alignAround(ctx: MindContext, q: Uint8Array, c: Uint8Array, qo: number, co: number): {
|
|
96
|
+
matched: Array<[number, number]>;
|
|
97
|
+
gaps: AlignGap[];
|
|
98
|
+
};
|
|
99
|
+
/** Contract a gap to its VARYING CORE: strip the prefix and suffix the two
|
|
100
|
+
* sides share. {@link alignAround} cannot match a shared affix shorter than
|
|
101
|
+
* W, so that affix lands INSIDE the gap — measured, the slot of
|
|
102
|
+
* `How do I compile main.c?` against `…hello.c?` comes back as
|
|
103
|
+
* `main.c?`/`hello.c?`, three bytes of which (`.c?`) both sides hold.
|
|
104
|
+
*
|
|
105
|
+
* Splicing the uncontracted gap carries the query's own punctuation into the
|
|
106
|
+
* answer; worse, it hides what actually VARIES, which is the only thing a
|
|
107
|
+
* cohort can agree about. Returns null when nothing is left on either side —
|
|
108
|
+
* a pure insertion or deletion, which names no slot. */
|
|
109
|
+
export declare function contractGap(q: Uint8Array, c: Uint8Array, g: AlignGap): AlignGap | null;
|
|
110
|
+
/** What one place two streams disagree IS, once contracted to its varying
|
|
111
|
+
* core. A consumer decides which kinds it can use; the matcher only reports.
|
|
112
|
+
*
|
|
113
|
+
* substitution both sides carry bytes — one thing stands where another does
|
|
114
|
+
* insertion the query carries bytes the candidate does not
|
|
115
|
+
* deletion the candidate carries bytes the query does not */
|
|
116
|
+
export type SlotKind = "substitution" | "insertion" | "deletion";
|
|
117
|
+
/** One VARIABLE POSITION of a pairing: where the query and a candidate differ,
|
|
118
|
+
* contracted to the bytes that actually vary. */
|
|
119
|
+
export interface FrameSlot {
|
|
120
|
+
/** Query span (empty for a deletion). */
|
|
121
|
+
qs: number;
|
|
122
|
+
qe: number;
|
|
123
|
+
/** Candidate span (empty for an insertion). */
|
|
124
|
+
cs: number;
|
|
125
|
+
ce: number;
|
|
126
|
+
kind: SlotKind;
|
|
127
|
+
/** The candidate's own bytes here — empty for an insertion. */
|
|
128
|
+
filler: Uint8Array;
|
|
129
|
+
}
|
|
130
|
+
/** One trained context read against the query as ONE structure with variable
|
|
131
|
+
* positions.
|
|
132
|
+
*
|
|
133
|
+
* EVERYTHING THE ALIGNER SAW, NOTHING JUDGED. `slots` holds every place the
|
|
134
|
+
* pairing varies, in query order, whatever its kind or size, and `covered`
|
|
135
|
+
* says how much of the query the two hold in common. No gate is applied
|
|
136
|
+
* here — see {@link frameSlots}. */
|
|
137
|
+
export interface FrameInstance {
|
|
138
|
+
/** The trained context this reading is against. */
|
|
139
|
+
id: number;
|
|
140
|
+
/** Every variable position, in query order. */
|
|
141
|
+
slots: FrameSlot[];
|
|
142
|
+
/** Query spans the pairing literally matched — the frame itself. */
|
|
143
|
+
matched: Array<[number, number]>;
|
|
144
|
+
/** Query bytes the frame accounts for: the size of what is shared. */
|
|
145
|
+
covered: number;
|
|
146
|
+
}
|
|
147
|
+
/** THE SLOT MATCHER: read one query ↔ context pairing as one structure with
|
|
148
|
+
* variable positions.
|
|
149
|
+
*
|
|
150
|
+
* IT REPORTS; IT DOES NOT JUDGE. This returns every gap the aligner found,
|
|
151
|
+
* contracted to its varying core and tagged with its kind, plus the shared
|
|
152
|
+
* coverage — and rejects nothing. That is the whole point of the split, and
|
|
153
|
+
* it was got WRONG first: four VOICING gates (the frame must dominate the
|
|
154
|
+
* query, each slot must reach one window on both sides, an insertion or
|
|
155
|
+
* deletion disqualifies the pairing, fillers must be pairwise distinct) were
|
|
156
|
+
* applied here, and every one of them is a requirement for SUBSTITUTING AND
|
|
157
|
+
* SPEAKING, not for knowing where a pairing varies. With them in place the
|
|
158
|
+
* shared reading was reference-shaped: measured over four real pairings, three
|
|
159
|
+
* were hidden from every consumer —
|
|
160
|
+
*
|
|
161
|
+
* `What is the capital of the country where the Eiffel Tower is?`
|
|
162
|
+
* against `What is the capital of France?` (covered 23/61) HIDDEN
|
|
163
|
+
* `What is the capital of France, really?` (an insertion) HIDDEN
|
|
164
|
+
* `What is the capital of Fran?` (sub-window) HIDDEN
|
|
165
|
+
*
|
|
166
|
+
* — including the case of the one consumer that most obviously needed it. A
|
|
167
|
+
* shared layer with one usable consumer is private code at a public address.
|
|
168
|
+
* Each gate now lives with the mechanism that needs it (see reference.ts).
|
|
169
|
+
*
|
|
170
|
+
* Seeded at the origin, because a frame is shared structure the query and its
|
|
171
|
+
* instances both OPEN with: the maximal run around (0,0) is the frame's head
|
|
172
|
+
* and the sweeps find the rest.
|
|
173
|
+
*
|
|
174
|
+
* Null only for a degenerate pairing (either side empty). */
|
|
175
|
+
export declare function frameSlots(ctx: MindContext, query: Uint8Array, cand: Uint8Array, id: number): FrameInstance | null;
|
|
176
|
+
/** Whether every member is byte-distinct from the others. */
|
|
177
|
+
export declare function distinct(items: readonly Uint8Array[]): boolean;
|
|
178
|
+
/** Substitute every `needle -> repl` pair SIMULTANEOUSLY: one left-to-right
|
|
179
|
+
* pass, longest needle first at each position, and a replacement is never
|
|
180
|
+
* re-examined.
|
|
181
|
+
*
|
|
182
|
+
* SIMULTANEOUS IS NOT A DETAIL. Applying the pairs in sequence lets one
|
|
183
|
+
* substitution's OUTPUT be another's input: with slots `gcc -> zig` and
|
|
184
|
+
* `hello.c -> zig.c` a sequential pass rewrites bytes it had just written,
|
|
185
|
+
* and the result depends on the order the slots happened to be found in.
|
|
186
|
+
* Longest-first at each position makes the pass independent of pair order,
|
|
187
|
+
* which is what keeps {@link carriesFillers} and the binding it licenses the
|
|
188
|
+
* SAME operation — if they could disagree, the licence would not be testing
|
|
189
|
+
* what is voiced. */
|
|
190
|
+
export declare function substituteAll(hay: Uint8Array, pairs: ReadonlyArray<{
|
|
191
|
+
needle: Uint8Array;
|
|
192
|
+
repl: Uint8Array;
|
|
193
|
+
}>): Uint8Array;
|
|
194
|
+
/** THE CARRIAGE LICENCE — the gate that decides whether a slot may be VOICED
|
|
195
|
+
* through. Given two instances of one frame and what each one continues to,
|
|
196
|
+
* it asks one byte question:
|
|
197
|
+
*
|
|
198
|
+
* substituteAll(contA, fillersA -> fillersB) == contB
|
|
199
|
+
*
|
|
200
|
+
* When it holds, the corpus attests byte-exactly that the continuation is a
|
|
201
|
+
* function of the fillers and nothing else, so putting a NEW occupant through
|
|
202
|
+
* the same carriage is derivation rather than invention. No threshold, no
|
|
203
|
+
* similarity, no new constant: the store's own instances decide, exactly as
|
|
204
|
+
* the bridge's `unanimous` decides whether a frame is a value slot.
|
|
205
|
+
*
|
|
206
|
+
* Its FAILURE is what this is really for. A frame whose continuation carries
|
|
207
|
+
* filler-DEPENDENT content — `What is the capital of X?` answering a different
|
|
208
|
+
* city per X — fails it, and that failure is the only thing between a slot
|
|
209
|
+
* and an invented fact. Measured on the trained 15.7M-node store (325,615
|
|
210
|
+
* contexts): `What is the capital of Zamunda?` resonates to a PURE cohort,
|
|
211
|
+
* every one of the top 14 hits an instance of that frame, with an unambiguous
|
|
212
|
+
* slot; every structural gate passes and only this one refuses, on
|
|
213
|
+
* `replace("Tokyo", "Japan" -> "France") != "Paris"`.
|
|
214
|
+
*
|
|
215
|
+
* With SEVERAL slots the test is unchanged, which is the point of testing the
|
|
216
|
+
* whole substitution at once: a frame whose answer tracks one slot but
|
|
217
|
+
* invents around another fails exactly as a single-slot value slot does. */
|
|
218
|
+
export declare function carriesFillers(contA: Uint8Array, fillersA: readonly Uint8Array[], contB: Uint8Array, fillersB: readonly Uint8Array[]): boolean;
|
|
60
219
|
/** The IN-LIST halo matcher: the best halo-mate for `halo` among EXPLICIT
|
|
61
220
|
* candidates, above the concept threshold — the list counterpart of
|
|
62
221
|
* {@link haloSiblings}, which asks the halo INDEX for candidates instead.
|
package/dist/src/mind/match.js
CHANGED
|
@@ -18,16 +18,21 @@
|
|
|
18
18
|
// direct or mutual-sibling)
|
|
19
19
|
// multi-hop pivot byte containment forward —
|
|
20
20
|
// articulation halo sibling substitute conceptThreshold
|
|
21
|
+
// reference frameSlots() (the shared carry into carriesFillers
|
|
22
|
+
// aligner, gaps contracted) the answer
|
|
21
23
|
//
|
|
22
24
|
// This module holds the shared vocabulary those configurations are built
|
|
23
|
-
// from — the MATCHERS (locate, alignRuns, alignGraded,
|
|
24
|
-
// the PROJECTIONS (follow, conceptHop, reverseContext,
|
|
25
|
-
// mechanism file states only its configuration, never its
|
|
26
|
-
// machinery.
|
|
25
|
+
// from — the MATCHERS (locate, alignRuns, alignGraded, alignAround/frameSlots,
|
|
26
|
+
// analogyStrength) and the PROJECTIONS (follow, conceptHop, reverseContext,
|
|
27
|
+
// project) — so each mechanism file states only its configuration, never its
|
|
28
|
+
// own copy of the machinery. Most gates live in geometry.ts (derived, never
|
|
29
|
+
// tuned); the two STRUCTURAL gates that are byte predicates rather than
|
|
30
|
+
// thresholds — isSpanShaped and carriesFillers — live here beside the matchers
|
|
31
|
+
// they gate.
|
|
27
32
|
import { addInto, cosine, dot, normalize, zeros } from "../vec.js";
|
|
28
|
-
import { conceptThreshold, identityBar, significanceBar } from "../geometry.js";
|
|
29
|
-
import { indexOf } from "../bytes.js";
|
|
30
|
-
import { leafIdRun } from "./canonical.js";
|
|
33
|
+
import { conceptThreshold, identityBar, significanceBar, } from "../geometry.js";
|
|
34
|
+
import { bytesEqual, indexOf } from "../bytes.js";
|
|
35
|
+
import { chainReach, leafIdRun } from "./canonical.js";
|
|
31
36
|
import { foldTree, gistOf, perceive, read, resolve } from "./primitives.js";
|
|
32
37
|
import { argmaxCosine, chooseAmong, chooseNext, corpusN, edgeAncestors, guidedFirst, hubBound, hubCap, sharedReachMemo, } from "./traverse.js";
|
|
33
38
|
import { recognise, segment } from "./recognition.js";
|
|
@@ -230,6 +235,294 @@ export function alignGraded(ctx, query, contextBytes, querySites) {
|
|
|
230
235
|
out.sort((a, b) => a.qs - b.qs);
|
|
231
236
|
return out;
|
|
232
237
|
}
|
|
238
|
+
/** Extend a seed match (query offset qo ↔ candidate offset co) to its maximal
|
|
239
|
+
* common run, then walk outward in both directions collecting further common
|
|
240
|
+
* runs of at least W bytes across bounded mismatch gaps (each side ≤
|
|
241
|
+
* chainReach). Returns the matched query spans and the mismatch pairs
|
|
242
|
+
* between consecutive runs.
|
|
243
|
+
*
|
|
244
|
+
* This is the SEEDED aligner, distinct from {@link alignRuns}: that one finds
|
|
245
|
+
* every run two structures share anywhere (a weave), this one reads two
|
|
246
|
+
* streams as ONE structure that diverges in bounded places (a frame with
|
|
247
|
+
* slots).
|
|
248
|
+
*
|
|
249
|
+
* Gaps come back in SWEEP order (right sweep, then left), not query order,
|
|
250
|
+
* and only the INTERIOR ones are reported — a consumer that needs the query's
|
|
251
|
+
* unmatched head or tail derives it from `matched`. Both are the bridge's
|
|
252
|
+
* contract, which prices its edges separately (see its matchStart/matchEnd
|
|
253
|
+
* window test); {@link frameSlots} takes the other reading. */
|
|
254
|
+
export function alignAround(ctx, q, c, qo, co) {
|
|
255
|
+
const W = ctx.space.maxGroup;
|
|
256
|
+
const reachCap = chainReach(W);
|
|
257
|
+
// Maximal run around the seed.
|
|
258
|
+
let qs = qo, ss = co;
|
|
259
|
+
while (qs > 0 && ss > 0 && q[qs - 1] === c[ss - 1]) {
|
|
260
|
+
qs--;
|
|
261
|
+
ss--;
|
|
262
|
+
}
|
|
263
|
+
let qe = qo, se = co;
|
|
264
|
+
while (qe < q.length && se < c.length && q[qe] === c[se]) {
|
|
265
|
+
qe++;
|
|
266
|
+
se++;
|
|
267
|
+
}
|
|
268
|
+
const matched = [[qs, qe]];
|
|
269
|
+
const gaps = [];
|
|
270
|
+
// The next common run of ≥ W bytes past (qi, si), with each side's gap
|
|
271
|
+
// bounded by chainReach; smallest total gap wins (nearest continuation).
|
|
272
|
+
const runLenAt = (qi, si) => {
|
|
273
|
+
let n = 0;
|
|
274
|
+
while (qi + n < q.length && si + n < c.length && q[qi + n] === c[si + n]) {
|
|
275
|
+
n++;
|
|
276
|
+
}
|
|
277
|
+
return n;
|
|
278
|
+
};
|
|
279
|
+
// RIGHT sweep.
|
|
280
|
+
let qi = qe, si = se;
|
|
281
|
+
for (;;) {
|
|
282
|
+
let found = false;
|
|
283
|
+
for (let total = 1; total <= 2 * reachCap && !found; total++) {
|
|
284
|
+
for (let gq = 0; gq <= Math.min(total, reachCap); gq++) {
|
|
285
|
+
const gs = total - gq;
|
|
286
|
+
if (gs > reachCap)
|
|
287
|
+
continue;
|
|
288
|
+
if (qi + gq >= q.length || si + gs >= c.length)
|
|
289
|
+
continue;
|
|
290
|
+
const n = runLenAt(qi + gq, si + gs);
|
|
291
|
+
if (n >= W || qi + gq + n === q.length) {
|
|
292
|
+
if (n === 0)
|
|
293
|
+
continue;
|
|
294
|
+
if (gq > 0 || gs > 0) {
|
|
295
|
+
gaps.push({ qs: qi, qe: qi + gq, cs: si, ce: si + gs });
|
|
296
|
+
}
|
|
297
|
+
matched.push([qi + gq, qi + gq + n]);
|
|
298
|
+
qi = qi + gq + n;
|
|
299
|
+
si = si + gs + n;
|
|
300
|
+
found = true;
|
|
301
|
+
break;
|
|
302
|
+
}
|
|
303
|
+
}
|
|
304
|
+
}
|
|
305
|
+
if (!found)
|
|
306
|
+
break;
|
|
307
|
+
}
|
|
308
|
+
// LEFT sweep (mirror).
|
|
309
|
+
qi = qs;
|
|
310
|
+
si = ss;
|
|
311
|
+
for (;;) {
|
|
312
|
+
let found = false;
|
|
313
|
+
for (let total = 1; total <= 2 * reachCap && !found; total++) {
|
|
314
|
+
for (let gq = 0; gq <= Math.min(total, reachCap); gq++) {
|
|
315
|
+
const gs = total - gq;
|
|
316
|
+
if (gs > reachCap)
|
|
317
|
+
continue;
|
|
318
|
+
if (qi - gq <= 0 || si - gs <= 0)
|
|
319
|
+
continue;
|
|
320
|
+
// Run ENDING at (qi - gq, si - gs).
|
|
321
|
+
let n = 0;
|
|
322
|
+
while (n < qi - gq && n < si - gs &&
|
|
323
|
+
q[qi - gq - 1 - n] === c[si - gs - 1 - n]) {
|
|
324
|
+
n++;
|
|
325
|
+
}
|
|
326
|
+
if (n >= W || n === qi - gq) {
|
|
327
|
+
if (n === 0)
|
|
328
|
+
continue;
|
|
329
|
+
if (gq > 0 || gs > 0) {
|
|
330
|
+
gaps.push({ qs: qi - gq, qe: qi, cs: si - gs, ce: si });
|
|
331
|
+
}
|
|
332
|
+
matched.push([qi - gq - n, qi - gq]);
|
|
333
|
+
qi = qi - gq - n;
|
|
334
|
+
si = si - gs - n;
|
|
335
|
+
found = true;
|
|
336
|
+
break;
|
|
337
|
+
}
|
|
338
|
+
}
|
|
339
|
+
}
|
|
340
|
+
if (!found)
|
|
341
|
+
break;
|
|
342
|
+
}
|
|
343
|
+
return { matched, gaps };
|
|
344
|
+
}
|
|
345
|
+
/** Contract a gap to its VARYING CORE: strip the prefix and suffix the two
|
|
346
|
+
* sides share. {@link alignAround} cannot match a shared affix shorter than
|
|
347
|
+
* W, so that affix lands INSIDE the gap — measured, the slot of
|
|
348
|
+
* `How do I compile main.c?` against `…hello.c?` comes back as
|
|
349
|
+
* `main.c?`/`hello.c?`, three bytes of which (`.c?`) both sides hold.
|
|
350
|
+
*
|
|
351
|
+
* Splicing the uncontracted gap carries the query's own punctuation into the
|
|
352
|
+
* answer; worse, it hides what actually VARIES, which is the only thing a
|
|
353
|
+
* cohort can agree about. Returns null when nothing is left on either side —
|
|
354
|
+
* a pure insertion or deletion, which names no slot. */
|
|
355
|
+
export function contractGap(q, c, g) {
|
|
356
|
+
let { qs, qe, cs, ce } = g;
|
|
357
|
+
while (qs < qe && cs < ce && q[qs] === c[cs]) {
|
|
358
|
+
qs++;
|
|
359
|
+
cs++;
|
|
360
|
+
}
|
|
361
|
+
while (qe > qs && ce > cs && q[qe - 1] === c[ce - 1]) {
|
|
362
|
+
qe--;
|
|
363
|
+
ce--;
|
|
364
|
+
}
|
|
365
|
+
return qe > qs && ce > cs ? { qs, qe, cs, ce } : null;
|
|
366
|
+
}
|
|
367
|
+
/** THE SLOT MATCHER: read one query ↔ context pairing as one structure with
|
|
368
|
+
* variable positions.
|
|
369
|
+
*
|
|
370
|
+
* IT REPORTS; IT DOES NOT JUDGE. This returns every gap the aligner found,
|
|
371
|
+
* contracted to its varying core and tagged with its kind, plus the shared
|
|
372
|
+
* coverage — and rejects nothing. That is the whole point of the split, and
|
|
373
|
+
* it was got WRONG first: four VOICING gates (the frame must dominate the
|
|
374
|
+
* query, each slot must reach one window on both sides, an insertion or
|
|
375
|
+
* deletion disqualifies the pairing, fillers must be pairwise distinct) were
|
|
376
|
+
* applied here, and every one of them is a requirement for SUBSTITUTING AND
|
|
377
|
+
* SPEAKING, not for knowing where a pairing varies. With them in place the
|
|
378
|
+
* shared reading was reference-shaped: measured over four real pairings, three
|
|
379
|
+
* were hidden from every consumer —
|
|
380
|
+
*
|
|
381
|
+
* `What is the capital of the country where the Eiffel Tower is?`
|
|
382
|
+
* against `What is the capital of France?` (covered 23/61) HIDDEN
|
|
383
|
+
* `What is the capital of France, really?` (an insertion) HIDDEN
|
|
384
|
+
* `What is the capital of Fran?` (sub-window) HIDDEN
|
|
385
|
+
*
|
|
386
|
+
* — including the case of the one consumer that most obviously needed it. A
|
|
387
|
+
* shared layer with one usable consumer is private code at a public address.
|
|
388
|
+
* Each gate now lives with the mechanism that needs it (see reference.ts).
|
|
389
|
+
*
|
|
390
|
+
* Seeded at the origin, because a frame is shared structure the query and its
|
|
391
|
+
* instances both OPEN with: the maximal run around (0,0) is the frame's head
|
|
392
|
+
* and the sweeps find the rest.
|
|
393
|
+
*
|
|
394
|
+
* Null only for a degenerate pairing (either side empty). */
|
|
395
|
+
export function frameSlots(ctx, query, cand, id) {
|
|
396
|
+
if (query.length === 0 || cand.length === 0)
|
|
397
|
+
return null;
|
|
398
|
+
const { matched, gaps } = alignAround(ctx, query, cand, 0, 0);
|
|
399
|
+
const spans = [...matched].sort((a, b) => a[0] - b[0]);
|
|
400
|
+
// Where the alignment RAN OUT on each side. Seeded at the origin there is
|
|
401
|
+
// no leading gap, so both cursors are everything consumed so far: the
|
|
402
|
+
// matched runs (equal length on both sides by construction) plus what each
|
|
403
|
+
// interior gap ate of its own side. Counting only the runs reads the
|
|
404
|
+
// candidate cursor short by exactly the fillers already seen, and invents a
|
|
405
|
+
// trailing gap on every well-aligned instance.
|
|
406
|
+
const all = [...gaps];
|
|
407
|
+
let qEnd = 0, cEnd = 0;
|
|
408
|
+
for (const [s, e] of spans) {
|
|
409
|
+
cEnd += e - s;
|
|
410
|
+
qEnd = Math.max(qEnd, e);
|
|
411
|
+
}
|
|
412
|
+
for (const g of gaps)
|
|
413
|
+
cEnd += g.ce - g.cs;
|
|
414
|
+
if (qEnd < query.length || cEnd < cand.length) {
|
|
415
|
+
all.push({ qs: qEnd, qe: query.length, cs: cEnd, ce: cand.length });
|
|
416
|
+
}
|
|
417
|
+
const slots = [];
|
|
418
|
+
for (const gap of all.sort((a, b) => a.qs - b.qs)) {
|
|
419
|
+
if (gap.qe <= gap.qs && gap.ce <= gap.cs)
|
|
420
|
+
continue;
|
|
421
|
+
// Contract to the varying core. contractGap returns null when one side is
|
|
422
|
+
// wholly shared with the other — a pure insertion or deletion, which is a
|
|
423
|
+
// real variation and is reported AS ONE, not discarded.
|
|
424
|
+
const core = contractGap(query, cand, gap);
|
|
425
|
+
const g = core ?? gap;
|
|
426
|
+
const kind = g.qe > g.qs && g.ce > g.cs
|
|
427
|
+
? "substitution"
|
|
428
|
+
: g.qe > g.qs
|
|
429
|
+
? "insertion"
|
|
430
|
+
: "deletion";
|
|
431
|
+
slots.push({
|
|
432
|
+
qs: g.qs,
|
|
433
|
+
qe: g.qe,
|
|
434
|
+
cs: g.cs,
|
|
435
|
+
ce: g.ce,
|
|
436
|
+
kind,
|
|
437
|
+
filler: cand.slice(g.cs, g.ce),
|
|
438
|
+
});
|
|
439
|
+
}
|
|
440
|
+
const covered = spans.reduce((n, [s, e]) => n + e - s, 0);
|
|
441
|
+
return { id, slots, matched: spans, covered };
|
|
442
|
+
}
|
|
443
|
+
/** Whether every member is byte-distinct from the others. */
|
|
444
|
+
export function distinct(items) {
|
|
445
|
+
for (let i = 0; i < items.length; i++) {
|
|
446
|
+
for (let j = i + 1; j < items.length; j++) {
|
|
447
|
+
if (bytesEqual(items[i], items[j]))
|
|
448
|
+
return false;
|
|
449
|
+
}
|
|
450
|
+
}
|
|
451
|
+
return true;
|
|
452
|
+
}
|
|
453
|
+
/** Substitute every `needle -> repl` pair SIMULTANEOUSLY: one left-to-right
|
|
454
|
+
* pass, longest needle first at each position, and a replacement is never
|
|
455
|
+
* re-examined.
|
|
456
|
+
*
|
|
457
|
+
* SIMULTANEOUS IS NOT A DETAIL. Applying the pairs in sequence lets one
|
|
458
|
+
* substitution's OUTPUT be another's input: with slots `gcc -> zig` and
|
|
459
|
+
* `hello.c -> zig.c` a sequential pass rewrites bytes it had just written,
|
|
460
|
+
* and the result depends on the order the slots happened to be found in.
|
|
461
|
+
* Longest-first at each position makes the pass independent of pair order,
|
|
462
|
+
* which is what keeps {@link carriesFillers} and the binding it licenses the
|
|
463
|
+
* SAME operation — if they could disagree, the licence would not be testing
|
|
464
|
+
* what is voiced. */
|
|
465
|
+
export function substituteAll(hay, pairs) {
|
|
466
|
+
const usable = pairs.filter((p) => p.needle.length > 0);
|
|
467
|
+
if (usable.length === 0)
|
|
468
|
+
return hay;
|
|
469
|
+
// Longest needle first, so a needle that is a prefix of another can never
|
|
470
|
+
// pre-empt it. Ties cannot arise: an instance whose fillers are not
|
|
471
|
+
// pairwise distinct is refused by frameSlots.
|
|
472
|
+
const order = [...usable].sort((a, b) => b.needle.length - a.needle.length);
|
|
473
|
+
const out = [];
|
|
474
|
+
let i = 0;
|
|
475
|
+
let hit = false;
|
|
476
|
+
outer: while (i < hay.length) {
|
|
477
|
+
for (const p of order) {
|
|
478
|
+
if (i + p.needle.length > hay.length)
|
|
479
|
+
continue;
|
|
480
|
+
let k = 0;
|
|
481
|
+
while (k < p.needle.length && hay[i + k] === p.needle[k])
|
|
482
|
+
k++;
|
|
483
|
+
if (k < p.needle.length)
|
|
484
|
+
continue;
|
|
485
|
+
for (const b of p.repl)
|
|
486
|
+
out.push(b);
|
|
487
|
+
i += p.needle.length;
|
|
488
|
+
hit = true;
|
|
489
|
+
continue outer;
|
|
490
|
+
}
|
|
491
|
+
out.push(hay[i]);
|
|
492
|
+
i++;
|
|
493
|
+
}
|
|
494
|
+
return hit ? Uint8Array.from(out) : hay;
|
|
495
|
+
}
|
|
496
|
+
/** THE CARRIAGE LICENCE — the gate that decides whether a slot may be VOICED
|
|
497
|
+
* through. Given two instances of one frame and what each one continues to,
|
|
498
|
+
* it asks one byte question:
|
|
499
|
+
*
|
|
500
|
+
* substituteAll(contA, fillersA -> fillersB) == contB
|
|
501
|
+
*
|
|
502
|
+
* When it holds, the corpus attests byte-exactly that the continuation is a
|
|
503
|
+
* function of the fillers and nothing else, so putting a NEW occupant through
|
|
504
|
+
* the same carriage is derivation rather than invention. No threshold, no
|
|
505
|
+
* similarity, no new constant: the store's own instances decide, exactly as
|
|
506
|
+
* the bridge's `unanimous` decides whether a frame is a value slot.
|
|
507
|
+
*
|
|
508
|
+
* Its FAILURE is what this is really for. A frame whose continuation carries
|
|
509
|
+
* filler-DEPENDENT content — `What is the capital of X?` answering a different
|
|
510
|
+
* city per X — fails it, and that failure is the only thing between a slot
|
|
511
|
+
* and an invented fact. Measured on the trained 15.7M-node store (325,615
|
|
512
|
+
* contexts): `What is the capital of Zamunda?` resonates to a PURE cohort,
|
|
513
|
+
* every one of the top 14 hits an instance of that frame, with an unambiguous
|
|
514
|
+
* slot; every structural gate passes and only this one refuses, on
|
|
515
|
+
* `replace("Tokyo", "Japan" -> "France") != "Paris"`.
|
|
516
|
+
*
|
|
517
|
+
* With SEVERAL slots the test is unchanged, which is the point of testing the
|
|
518
|
+
* whole substitution at once: a frame whose answer tracks one slot but
|
|
519
|
+
* invents around another fails exactly as a single-slot value slot does. */
|
|
520
|
+
export function carriesFillers(contA, fillersA, contB, fillersB) {
|
|
521
|
+
if (fillersA.length !== fillersB.length)
|
|
522
|
+
return false;
|
|
523
|
+
const projected = substituteAll(contA, fillersA.map((needle, s) => ({ needle, repl: fillersB[s] })));
|
|
524
|
+
return bytesEqual(projected, contB);
|
|
525
|
+
}
|
|
233
526
|
/** The IN-LIST halo matcher: the best halo-mate for `halo` among EXPLICIT
|
|
234
527
|
* candidates, above the concept threshold — the list counterpart of
|
|
235
528
|
* {@link haloSiblings}, which asks the halo INDEX for candidates instead.
|
|
@@ -0,0 +1,22 @@
|
|
|
1
|
+
import type { MindContext } from "../types.js";
|
|
2
|
+
import type { PipelineMechanism } from "../pipeline-mechanism.js";
|
|
3
|
+
/** A trained form the query opens, and the bytes by which it continues. */
|
|
4
|
+
export interface PrefixCompletion {
|
|
5
|
+
/** The trained form whose opening the query is — the answer, voiced whole. */
|
|
6
|
+
id: number;
|
|
7
|
+
/** The form's own bytes. The mechanism grounds a FORM, never a slice of
|
|
8
|
+
* one: slicing at the query's end would cut at an offset the geometry has
|
|
9
|
+
* no reason to treat as a boundary. */
|
|
10
|
+
form: Uint8Array;
|
|
11
|
+
/** The bytes past the query — carried for the rationale and for the
|
|
12
|
+
* uniqueness comparison, not voiced on its own. */
|
|
13
|
+
continuation: Uint8Array;
|
|
14
|
+
}
|
|
15
|
+
/** The sole trained form the query opens — or null when no candidate opens with
|
|
16
|
+
* it, when the continuation is sub-quantum, when a candidate's continuation
|
|
17
|
+
* cannot be read through, or when the candidates disagree.
|
|
18
|
+
*
|
|
19
|
+
* `ranked` must be a list the caller has ALREADY fetched; this mechanism never
|
|
20
|
+
* resonates on its own (see the header's cost note). */
|
|
21
|
+
export declare function prefixCompletion(ctx: MindContext, query: Uint8Array, ranked: ReadonlyArray<number>): PrefixCompletion | null;
|
|
22
|
+
export declare const prefixMechanism: PipelineMechanism;
|