@polycode-projects/the-mechanical-code-talker 2.11.11 → 3.0.0
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/README.md +4 -2
- package/bin/tmct.mjs +33 -2
- package/corpus/tier2/generate.mjs +1 -1
- package/package.json +3 -2
- package/src/adapters/source.mjs +8 -6
- package/src/adapters/toml-config.mjs +3 -2
- package/src/domain/cli-verbs.mjs +9 -0
- package/src/domain/grammar/ace.mjs +16 -1
- package/src/index/extract-jsts.mjs +251 -0
- package/src/index/extract-python.mjs +46 -0
- package/src/index/extract_ast.py +364 -0
- package/src/index/index-repo.mjs +173 -0
- package/src/index/registry.mjs +37 -0
- package/src/index/spawn.mjs +35 -0
- package/src/index/walk.mjs +0 -0
- package/src/services/adventure-viz.mjs +23 -1
- package/src/services/adventure.mjs +69 -3
- package/src/services/chat-session.mjs +3 -3
- package/src/services/chat.mjs +12 -10
- package/src/services/research-viz.mjs +34 -6
- package/src/services/research.mjs +154 -44
- package/src/surfaces/web/memory-ask-browser.bundle.js +72 -72
- package/src/surfaces/web/research-browser-entry.mjs +12 -1
|
@@ -9,7 +9,7 @@
|
|
|
9
9
|
// share the slot.
|
|
10
10
|
|
|
11
11
|
import { worldProvenanceTag } from "../domain/worlds-pack.mjs";
|
|
12
|
-
import { parseImperative } from "../domain/grammar/ace.mjs";
|
|
12
|
+
import { parseImperative, OBJECT_PRONOUNS } from "../domain/grammar/ace.mjs";
|
|
13
13
|
import { createCompletionsGraphAdapter } from "../domain/completions/graph-adapter.mjs";
|
|
14
14
|
import { actionFamilies } from "../domain/router/taught.mjs";
|
|
15
15
|
import { compileDomain, precondHolds, roleBinding } from "../domain/domain.mjs";
|
|
@@ -1224,6 +1224,63 @@ function renderedImperativeCommand(cmd) {
|
|
|
1224
1224
|
return parts.join(" ");
|
|
1225
1225
|
}
|
|
1226
1226
|
|
|
1227
|
+
// ---- pronoun binding: the session focus ---------------------------------------
|
|
1228
|
+
//
|
|
1229
|
+
// A world command may name its object with a pronoun ("examine it", "take
|
|
1230
|
+
// them", "talk to him") instead of a noun. The antecedent is not in the
|
|
1231
|
+
// sentence — it's the last thing the player successfully acted on this
|
|
1232
|
+
// session, the FOCUS — so the parser leaves the pronoun bare (ace.mjs's
|
|
1233
|
+
// OBJECT_PRONOUNS) and the lane binds it here, through ONE seam that every
|
|
1234
|
+
// object-taking verb passes on its way to runWorldCommand. With no focus
|
|
1235
|
+
// standing, a pronoun gets an honest reference nudge, never the vocabulary
|
|
1236
|
+
// decline (a pronoun is a reference, not an unknown word).
|
|
1237
|
+
|
|
1238
|
+
const PRONOUN_SLOTS = ["object", "indirectObject", "instrument"];
|
|
1239
|
+
|
|
1240
|
+
const commandHasPronoun = (cmd) => PRONOUN_SLOTS.some((s) => cmd[s] && OBJECT_PRONOUNS.has(cmd[s]));
|
|
1241
|
+
|
|
1242
|
+
/** A pronoun command with no focus standing: the reference nudge, embedding a
|
|
1243
|
+
* real, actionable object from the current room when one is on show (else a
|
|
1244
|
+
* static example). Never the "I don't know the word" line — the vocabulary
|
|
1245
|
+
* misdiagnosis is unreachable for a pronoun. */
|
|
1246
|
+
async function noFocusPronounNudge(pronoun, { memoryDir }) {
|
|
1247
|
+
let example = null;
|
|
1248
|
+
try {
|
|
1249
|
+
const rows = readFactRows(await loadMemory(memoryDir));
|
|
1250
|
+
const state = foldWorldState(worldActionRows(rows));
|
|
1251
|
+
const here = state.placements.get("player")?.object ?? null;
|
|
1252
|
+
if (here) {
|
|
1253
|
+
for (const action of roomAffordances(rows, state, here)) {
|
|
1254
|
+
const m = action.match(/^(?:examine|take|open|unlock|talk to) (.+)$/);
|
|
1255
|
+
if (m) { example = m[1]; break; }
|
|
1256
|
+
}
|
|
1257
|
+
}
|
|
1258
|
+
} catch { /* no probe available — the static example carries the nudge */ }
|
|
1259
|
+
const eg = example ?? "lamp";
|
|
1260
|
+
return answer(
|
|
1261
|
+
`I'm not sure what "${pronoun}" refers to yet — name the thing, e.g. "examine ${eg}".`,
|
|
1262
|
+
`ADVENTURE — pronoun "${pronoun}" arrived with no focus standing; asked which thing it means, never the vocabulary decline`,
|
|
1263
|
+
{ miss: true },
|
|
1264
|
+
);
|
|
1265
|
+
}
|
|
1266
|
+
|
|
1267
|
+
/** Bind any pronoun object/indirect/instrument slot to the session focus.
|
|
1268
|
+
* Returns `{ cmd }` with the pronouns rewritten to the focus term, or `{
|
|
1269
|
+
* nudge }` (the reference nudge) when a pronoun stands but no focus does. A
|
|
1270
|
+
* command with no pronoun passes straight through untouched. */
|
|
1271
|
+
async function bindPronouns(cmd, { focus, memoryDir }) {
|
|
1272
|
+
if (!commandHasPronoun(cmd)) return { cmd };
|
|
1273
|
+
if (!focus) {
|
|
1274
|
+
const pronoun = PRONOUN_SLOTS.map((s) => cmd[s]).find((v) => v && OBJECT_PRONOUNS.has(v));
|
|
1275
|
+
return { nudge: await noFocusPronounNudge(pronoun, { memoryDir }) };
|
|
1276
|
+
}
|
|
1277
|
+
const bound = { ...cmd };
|
|
1278
|
+
for (const s of PRONOUN_SLOTS) {
|
|
1279
|
+
if (bound[s] && OBJECT_PRONOUNS.has(bound[s])) bound[s] = focus;
|
|
1280
|
+
}
|
|
1281
|
+
return { cmd: bound };
|
|
1282
|
+
}
|
|
1283
|
+
|
|
1227
1284
|
// ---- the lane ----------------------------------------------------------------
|
|
1228
1285
|
|
|
1229
1286
|
/**
|
|
@@ -1292,9 +1349,18 @@ export async function adventureTurn(line, { planHolder, memoryDir, sessionId = "
|
|
|
1292
1349
|
};
|
|
1293
1350
|
}
|
|
1294
1351
|
if (INVENTORY_RE.test(line)) return inventoryAnswer({ memoryDir, graph });
|
|
1295
|
-
const
|
|
1296
|
-
if (
|
|
1352
|
+
const parsed = parseImperative(line, lexicon ?? undefined);
|
|
1353
|
+
if (parsed) {
|
|
1354
|
+
const bound = await bindPronouns(parsed, { focus: adventure.focus, memoryDir });
|
|
1355
|
+
if (bound.nudge) return bound.nudge;
|
|
1356
|
+
const cmd = bound.cmd;
|
|
1297
1357
|
const result = await runWorldCommand(cmd, { world: adventure.world, memoryDir, env, graph, cache });
|
|
1358
|
+
// The object a command SUCCESSFULLY named becomes the focus a later
|
|
1359
|
+
// pronoun binds to — so "look lamp" then "examine it" reads the lamp, and
|
|
1360
|
+
// "talk to housekeeper" makes "him"/"her" the housekeeper. A miss leaves
|
|
1361
|
+
// the standing focus untouched; a bare room look or a move carries no
|
|
1362
|
+
// object and so never disturbs it.
|
|
1363
|
+
if (!result.miss && cmd.object) adventure.focus = cmd.object;
|
|
1298
1364
|
if (!cmd.corrected?.length) return result;
|
|
1299
1365
|
// A fuzzy-repaired verb or direction still executes normally, but the
|
|
1300
1366
|
// response says what it read the line as, so a genuine miss is never
|
|
@@ -332,9 +332,9 @@ export async function createSession({
|
|
|
332
332
|
// no code graph → point at how to GET one (a graph producer / --repo / the shipped
|
|
333
333
|
// example), and at what IS answerable now — `vocabHint` is only ever a term
|
|
334
334
|
// confirmed to resolve in THIS session's actual seed state (see vocabExampleHint),
|
|
335
|
-
// never a hardcoded example that might not have been seeded. tmct
|
|
336
|
-
//
|
|
337
|
-
...(noCodeGraph ? [`for code structure, point me at a .tmct/graph.json with --repo <path> or try \`npm run example:mini\`
|
|
335
|
+
// never a hardcoded example that might not have been seeded. tmct can index a
|
|
336
|
+
// repo itself (`tmct index`) or read a graph any other producer wrote.
|
|
337
|
+
...(noCodeGraph ? [`for code structure, index this repo with \`tmct index\`, or point me at a .tmct/graph.json with --repo <path> (or try \`npm run example:mini\`). ${vocabHint}`] : []),
|
|
338
338
|
"pass --repo <path> to target a different repo",
|
|
339
339
|
"ask a question, or /help for commands (/stats for an overview) — /exit to leave",
|
|
340
340
|
];
|
package/src/services/chat.mjs
CHANGED
|
@@ -602,7 +602,7 @@ export function answerCount(graph, query) {
|
|
|
602
602
|
// empty — an honest, non-dangling message pointing at how to load one.
|
|
603
603
|
if (!kinds.length) {
|
|
604
604
|
return `I can't count "${noun}" — no code graph is loaded yet, so there's nothing to count ` +
|
|
605
|
-
`(point me at
|
|
605
|
+
`(index this repo with "tmct index", point me at another with --repo, or run "npm run example:mini").`;
|
|
606
606
|
}
|
|
607
607
|
return `I can't count "${noun}". I count: ${kinds.join(", ")}. ` +
|
|
608
608
|
`Try "how many classes are there".`;
|
|
@@ -2077,8 +2077,8 @@ function orientationAnswer(templates, graph, vocabHint) {
|
|
|
2077
2077
|
* null), matching the file's "never crash, always degrade to one honest line"
|
|
2078
2078
|
* ethos. Kept short and hand-written so it never drifts silently. */
|
|
2079
2079
|
const ORIENTATION_EMPTY_FALLBACK = "I'm tmct — a deterministic, offline chat assistant (no LLM). "
|
|
2080
|
-
+ "For code structure (imports, calls, definitions) point me at a repo with `--repo <path>`, "
|
|
2081
|
-
+ "or try the shipped example `npm run example:mini`.
|
|
2080
|
+
+ "For code structure (imports, calls, definitions) run `tmct index` here, point me at a repo with `--repo <path>`, "
|
|
2081
|
+
+ "or try the shipped example `npm run example:mini`. /help for commands.";
|
|
2082
2082
|
|
|
2083
2083
|
/** A dynamic orientation string for the meta/self lane: a /stats-style overview
|
|
2084
2084
|
* when a code graph is loaded, else the honest empty-graph orientation — rendered
|
|
@@ -5974,7 +5974,7 @@ export async function helpText() {
|
|
|
5974
5974
|
["/ingest <path>", "read a local text file and store every fact the recognizer grounds from it (same recognizer as `tmct extract`)"],
|
|
5975
5975
|
["/narrate on|off", "verbose developer/debug mode: decision points, matched pattern, results+sources, goal per turn"],
|
|
5976
5976
|
["/wiki on|off|supplement|always", "live Wikipedia (default off): on tries en.wikipedia.org when I can't answer (network), cited; supplement also adds a read-out under every grounded vocabulary answer; always widens that to every grounded answer"],
|
|
5977
|
-
["research <topic> [limit N]", "fetch the topic from Simple English Wikipedia (the explicit ask is the network consent), store what it grounds, and queue its linked topics — \"research next\" steps the queue; also status/stop"],
|
|
5977
|
+
["research <topic> [limit N] [depth D]", "fetch the topic from Simple English Wikipedia (the explicit ask is the network consent), store what it grounds, and queue its linked topics — \"research next\" steps the queue; also status/stop. limit N caps the links queued per topic, depth D how many hops the queue follows (1 by default); a run also stops at its total node budget"],
|
|
5978
5978
|
["/help", "this list"],
|
|
5979
5979
|
["/exit", "leave the session (also Ctrl+C / Ctrl+D)"],
|
|
5980
5980
|
];
|
|
@@ -13009,8 +13009,8 @@ async function runAsk(query, { config, source, graph, focus, last, templates, me
|
|
|
13009
13009
|
answer = `${answer}\n(I don't know that yet — you can teach me: say "remember: <thing> is a <kind>".)`;
|
|
13010
13010
|
note(trace, "intermediate: HONEST-EMPTY POLISH — a browser miss points at the teach lane, not the CLI-only --repo remedy");
|
|
13011
13011
|
} else {
|
|
13012
|
-
answer = `${answer}\n(this repo has no code graph —
|
|
13013
|
-
note(trace, "intermediate: HONEST-EMPTY POLISH — the loaded graph has 0 modules, so the dead-end got a
|
|
13012
|
+
answer = `${answer}\n(this repo has no code graph — index it with \`tmct index\`, point me at a \`.tmct/graph.json\` with \`--repo <path>\`, or run \`npm run example:mini\`.)`;
|
|
13013
|
+
note(trace, "intermediate: HONEST-EMPTY POLISH — the loaded graph has 0 modules, so the dead-end got a tmct index/--repo pointer appended");
|
|
13014
13014
|
}
|
|
13015
13015
|
}
|
|
13016
13016
|
// TEACH-OFFER: a "what is X" miss where X is genuinely unknown EVERYWHERE —
|
|
@@ -14540,10 +14540,12 @@ export async function runTurn(input, { config, source = defaultSource, graph = n
|
|
|
14540
14540
|
}
|
|
14541
14541
|
}
|
|
14542
14542
|
|
|
14543
|
-
// RESEARCH — "research <topic>[, limit N]" runs a Simple English
|
|
14544
|
-
// queue through the same ingest path a live-Wikipedia rescue uses:
|
|
14545
|
-
// now, the lead section's linked topics queued for "research next"
|
|
14546
|
-
// the web pages' auto-play button submits turn by turn)
|
|
14543
|
+
// RESEARCH — "research <topic>[, limit N][, depth D]" runs a Simple English
|
|
14544
|
+
// Wikipedia queue through the same ingest path a live-Wikipedia rescue uses:
|
|
14545
|
+
// depth 0 now, the lead section's linked topics queued for "research next"
|
|
14546
|
+
// (which the web pages' auto-play button submits turn by turn), and each of
|
|
14547
|
+
// those fanning out again while the run's depth knob allows, up to its total
|
|
14548
|
+
// node budget. The explicit
|
|
14547
14549
|
// request is the network consent for its own fetches — unlike the
|
|
14548
14550
|
// clean-miss rescue, which fires on an ordinary question and so stays
|
|
14549
14551
|
// behind /wiki on. Queue state threads turn-to-turn as researchState, the
|
|
@@ -128,6 +128,9 @@ ${THEME_TOKENS_CSS}
|
|
|
128
128
|
.card .note { font-family: ${MONO_STACK}; font-size: .68rem; color: var(--muted); min-height: 1rem; }
|
|
129
129
|
.optionToggle { display: inline-flex; align-items: center; gap: .35rem; font-family: ${MONO_STACK}; font-size: .68rem; color: var(--muted); cursor: pointer; }
|
|
130
130
|
.optionToggle input { margin: 0; accent-color: var(--corpus); }
|
|
131
|
+
.knobs { display: flex; gap: .9rem; align-items: center; flex-wrap: wrap; }
|
|
132
|
+
.knob { display: inline-flex; align-items: center; gap: .4rem; font-family: ${MONO_STACK}; font-size: .68rem; color: var(--muted); }
|
|
133
|
+
.knob input[type="number"] { width: 3.4rem; font-family: ${MONO_STACK}; font-size: .74rem; background: var(--card); color: var(--ink); border: 1px solid var(--line); border-radius: 6px; padding: .25rem .35rem; text-align: right; }
|
|
131
134
|
|
|
132
135
|
/* highlights + ask, two columns */
|
|
133
136
|
.cols { display: grid; grid-template-columns: 1fr 1fr; gap: 1.4rem; align-items: start; }
|
|
@@ -197,6 +200,14 @@ ${THEME_TOKENS_CSS}
|
|
|
197
200
|
<button type="button" class="btn" id="researchNext" hidden>research next</button>
|
|
198
201
|
<button type="button" class="btn" id="researchPlay" aria-pressed="false" hidden>play</button>
|
|
199
202
|
</div>
|
|
203
|
+
<div class="knobs">
|
|
204
|
+
<label class="knob" title="How many topics one research run may fetch and store in total (the first topic counts as one). A deep run stops fetching once it reaches this budget.">
|
|
205
|
+
max nodes <input id="researchNodes" type="number" min="1" max="50" step="1" value="12" inputmode="numeric" aria-label="Maximum response nodes">
|
|
206
|
+
</label>
|
|
207
|
+
<label class="knob" title="How deep the link fan-out follows: depth 1 is the topic's own lead links, depth 2 those topics' links, and so on. Applies to the next run you start.">
|
|
208
|
+
max depth <input id="researchDepth" type="number" min="1" max="3" step="1" value="1" inputmode="numeric" aria-label="Maximum node depth">
|
|
209
|
+
</label>
|
|
210
|
+
</div>
|
|
200
211
|
<p class="note" id="researchNote"></p>
|
|
201
212
|
</div>
|
|
202
213
|
<div class="card">
|
|
@@ -593,15 +604,31 @@ ${THEME_TOKENS_CSS}
|
|
|
593
604
|
el("researchPlay").setAttribute("aria-pressed", String(state.playing));
|
|
594
605
|
const note = el("researchNote");
|
|
595
606
|
if (!researchQueue) { /* leave whatever the last turn's note said */ }
|
|
596
|
-
else
|
|
597
|
-
|
|
598
|
-
|
|
599
|
-
|
|
600
|
-
|
|
601
|
-
|
|
607
|
+
else {
|
|
608
|
+
const depth = researchQueue.maxDepth || 1;
|
|
609
|
+
const budget = researchQueue.maxTopics || 0;
|
|
610
|
+
const knobs = " (depth " + depth + (budget ? ", budget " + budget : "") + ")";
|
|
611
|
+
const capped = researchQueue.nodeCapReached ? " — node budget reached" : "";
|
|
612
|
+
if (researchQueue.complete) {
|
|
613
|
+
note.textContent = 'research "' + researchQueue.topic + '" complete — '
|
|
614
|
+
+ researchQueue.done.length + " topic" + (researchQueue.done.length === 1 ? "" : "s") + " grounded" + knobs + capped + ".";
|
|
615
|
+
} else {
|
|
616
|
+
note.textContent = 'research "' + researchQueue.topic + '": '
|
|
617
|
+
+ researchQueue.done.length + " done · " + researchQueue.pending.length + " queued" + knobs + capped + ".";
|
|
618
|
+
}
|
|
602
619
|
}
|
|
603
620
|
}
|
|
604
621
|
|
|
622
|
+
// Read the two node knobs off the page and hand them to the session for the
|
|
623
|
+
// NEXT run started. A run already going keeps the knobs it captured.
|
|
624
|
+
function applyResearchConfig() {
|
|
625
|
+
if (!session || !session.setResearchConfig) return;
|
|
626
|
+
session.setResearchConfig({
|
|
627
|
+
maxTopics: parseInt(el("researchNodes").value, 10),
|
|
628
|
+
maxDepth: parseInt(el("researchDepth").value, 10),
|
|
629
|
+
});
|
|
630
|
+
}
|
|
631
|
+
|
|
605
632
|
async function researchStep(line) {
|
|
606
633
|
if (!session) return;
|
|
607
634
|
let res;
|
|
@@ -619,6 +646,7 @@ ${THEME_TOKENS_CSS}
|
|
|
619
646
|
async function startResearch() {
|
|
620
647
|
const topic = el("researchTopic").value.trim();
|
|
621
648
|
if (!topic || !session) return;
|
|
649
|
+
applyResearchConfig();
|
|
622
650
|
el("researchTopic").value = "";
|
|
623
651
|
el("researchNote").textContent = 'researching "' + topic + '"…';
|
|
624
652
|
const previous = researchQueue;
|
|
@@ -42,34 +42,64 @@ export function researchTopicKey(topic, lexicon = null) {
|
|
|
42
42
|
return t;
|
|
43
43
|
}
|
|
44
44
|
|
|
45
|
-
/** The most linked topics any
|
|
46
|
-
* the
|
|
45
|
+
/** The most linked topics any single fan-out may queue — the per-fan-out cap
|
|
46
|
+
* the request's "limit N" (or the configured `fanoutLimit`) sets, itself
|
|
47
|
+
* bounded here. */
|
|
47
48
|
export const RESEARCH_FANOUT_MAX = 12;
|
|
48
49
|
|
|
50
|
+
/** How deep the link fan-out may follow: depth 0 is the requested topic, and
|
|
51
|
+
* each further tier is the previous tier's own lead-section links. The user
|
|
52
|
+
* knob (page "maximum node depth", CLI `depth D`) is clamped to this. */
|
|
53
|
+
export const RESEARCH_MAX_DEPTH = 3;
|
|
54
|
+
|
|
55
|
+
/** The largest total node budget a run may carry — the page's "maximum
|
|
56
|
+
* response nodes" upper bound. `maxTopics` caps how many topics one run
|
|
57
|
+
* fetches and stores in total (depth 0 counts as the first). */
|
|
58
|
+
export const RESEARCH_MAX_TOPICS = 50;
|
|
59
|
+
|
|
49
60
|
export const RESEARCH_DEFAULTS = Object.freeze({
|
|
50
61
|
fanoutLimit: 5,
|
|
51
|
-
|
|
62
|
+
maxDepth: 1,
|
|
63
|
+
maxTopics: 12,
|
|
52
64
|
minIntervalMs: 2000,
|
|
53
65
|
});
|
|
54
66
|
|
|
55
67
|
const clampInt = (n, lo, hi) => Math.min(hi, Math.max(lo, Math.floor(n)));
|
|
56
68
|
|
|
69
|
+
/** A partial `{ fanoutLimit?, maxDepth?, maxTopics?, minIntervalMs? }` (camelCase,
|
|
70
|
+
* as the page and CLI supply it) folded onto the shipped defaults and clamped
|
|
71
|
+
* to the engineered ranges: fan-out at RESEARCH_FANOUT_MAX, depth at
|
|
72
|
+
* RESEARCH_MAX_DEPTH, the node budget at [1, RESEARCH_MAX_TOPICS], and the
|
|
73
|
+
* polite interval only ever RAISED above its floor, never lowered. Every
|
|
74
|
+
* non-finite field falls back to its default, so a corrupt/absent value is
|
|
75
|
+
* the shipped knob, never a crash. */
|
|
76
|
+
export function clampResearchConfig(partial = {}) {
|
|
77
|
+
const cfg = { ...RESEARCH_DEFAULTS };
|
|
78
|
+
const fanout = Number(partial.fanoutLimit);
|
|
79
|
+
if (Number.isFinite(fanout)) cfg.fanoutLimit = clampInt(fanout, 0, RESEARCH_FANOUT_MAX);
|
|
80
|
+
const depth = Number(partial.maxDepth);
|
|
81
|
+
if (Number.isFinite(depth)) cfg.maxDepth = clampInt(depth, 0, RESEARCH_MAX_DEPTH);
|
|
82
|
+
const topics = Number(partial.maxTopics);
|
|
83
|
+
if (Number.isFinite(topics)) cfg.maxTopics = clampInt(topics, 1, RESEARCH_MAX_TOPICS);
|
|
84
|
+
const interval = Number(partial.minIntervalMs);
|
|
85
|
+
if (Number.isFinite(interval)) cfg.minIntervalMs = Math.max(RESEARCH_DEFAULTS.minIntervalMs, interval);
|
|
86
|
+
return cfg;
|
|
87
|
+
}
|
|
88
|
+
|
|
57
89
|
/** tmct.toml's `[research]` table → the lane's effective knobs, shipped
|
|
58
90
|
* defaults filling every unset key (the same posture resolveGameConfig
|
|
59
91
|
* takes with `[games.*]`). `fanout_limit` caps at RESEARCH_FANOUT_MAX;
|
|
60
|
-
* `depth_limit`
|
|
61
|
-
* `
|
|
62
|
-
* never lower it. */
|
|
92
|
+
* `depth_limit`/`max_depth` set how deep the fan-out follows (0 means no
|
|
93
|
+
* fan-out); `max_topics` sets the total node budget; `min_interval_ms` may
|
|
94
|
+
* only RAISE the polite floor between round trips, never lower it. */
|
|
63
95
|
export function resolveResearchConfig(toml = null) {
|
|
64
96
|
const raw = toml?.research || {};
|
|
65
|
-
|
|
66
|
-
|
|
67
|
-
|
|
68
|
-
|
|
69
|
-
|
|
70
|
-
|
|
71
|
-
if (Number.isFinite(interval)) cfg.minIntervalMs = Math.max(RESEARCH_DEFAULTS.minIntervalMs, interval);
|
|
72
|
-
return cfg;
|
|
97
|
+
return clampResearchConfig({
|
|
98
|
+
fanoutLimit: raw.fanout_limit,
|
|
99
|
+
maxDepth: raw.max_depth ?? raw.depth_limit,
|
|
100
|
+
maxTopics: raw.max_topics,
|
|
101
|
+
minIntervalMs: raw.min_interval_ms,
|
|
102
|
+
});
|
|
73
103
|
}
|
|
74
104
|
|
|
75
105
|
// The verbs that step/inspect/end a run, checked before the start shape so
|
|
@@ -77,16 +107,21 @@ export function resolveResearchConfig(toml = null) {
|
|
|
77
107
|
const RESEARCH_NEXT_RE = /^research[,:]?\s+(?:next|continue|more)\s*[.!?]*$/i;
|
|
78
108
|
const RESEARCH_STATUS_RE = /^research[,:]?\s+status\s*[.!?]*$/i;
|
|
79
109
|
const RESEARCH_STOP_RE = /^research[,:]?\s+(?:stop|cancel|quit|end)\s*[.!?]*$/i;
|
|
80
|
-
const RESEARCH_START_RE = /^research[,:]?\s+(.+?)
|
|
110
|
+
const RESEARCH_START_RE = /^research[,:]?\s+(.+?)\s*[.!?]*$/i;
|
|
111
|
+
// The trailing knob tokens a start request may carry, stripped one at a time
|
|
112
|
+
// off the END so "limit N" and "depth D" read in either order: "research owls,
|
|
113
|
+
// limit 2 depth 2" and "research owls depth 2, limit 2" both parse the same.
|
|
114
|
+
const RESEARCH_OPTION_RE = /[,;]?\s+(?:with\s+)?(limit|depth)\s+(\d{1,3})$/i;
|
|
81
115
|
// A bare continuation word steps the queue too, but only when a run is
|
|
82
116
|
// actually pending and no plan lane owns the word — parseResearchRequest
|
|
83
117
|
// reports it as its own kind so the caller can apply that gate.
|
|
84
118
|
const BARE_NEXT_RE = /^(?:next|continue|carry on|keep going)\s*[.!?]*$/i;
|
|
85
119
|
|
|
86
|
-
/** The research request a line carries, or null. Kinds: start {topic,
|
|
87
|
-
*
|
|
88
|
-
*
|
|
89
|
-
* present when the request
|
|
120
|
+
/** The research request a line carries, or null. Kinds: start {topic, limit?,
|
|
121
|
+
* depth?}, next, bareNext, status, stop. The topic keeps the user's own words
|
|
122
|
+
* minus a leading article and any wrapping quotes; `limit` (per-fan-out cap)
|
|
123
|
+
* and `depth` (how deep the fan-out follows) are present only when the request
|
|
124
|
+
* named them. */
|
|
90
125
|
export function parseResearchRequest(line) {
|
|
91
126
|
const q = String(line || "").trim();
|
|
92
127
|
if (!q) return null;
|
|
@@ -96,13 +131,21 @@ export function parseResearchRequest(line) {
|
|
|
96
131
|
if (RESEARCH_STOP_RE.test(q)) return { kind: "stop" };
|
|
97
132
|
const m = q.match(RESEARCH_START_RE);
|
|
98
133
|
if (!m) return null;
|
|
99
|
-
|
|
134
|
+
let rest = m[1].trim();
|
|
135
|
+
const opts = {};
|
|
136
|
+
for (let om = rest.match(RESEARCH_OPTION_RE); om; om = rest.match(RESEARCH_OPTION_RE)) {
|
|
137
|
+
const kind = om[1].toLowerCase();
|
|
138
|
+
if (opts[kind] === undefined) opts[kind] = Number(om[2]);
|
|
139
|
+
rest = rest.slice(0, om.index).trim();
|
|
140
|
+
}
|
|
141
|
+
const topic = rest
|
|
100
142
|
.replace(/^["'‘’“”]+|["'‘’“”]+$/g, "")
|
|
101
143
|
.replace(/^(?:an?|the)\s+/i, "")
|
|
102
144
|
.trim();
|
|
103
145
|
if (!topic) return null;
|
|
104
146
|
const out = { kind: "start", topic };
|
|
105
|
-
if (
|
|
147
|
+
if (opts.limit !== undefined) out.limit = opts.limit;
|
|
148
|
+
if (opts.depth !== undefined) out.depth = opts.depth;
|
|
106
149
|
return out;
|
|
107
150
|
}
|
|
108
151
|
|
|
@@ -122,29 +165,101 @@ export function renderResearchAnswer(term, article) {
|
|
|
122
165
|
}
|
|
123
166
|
|
|
124
167
|
/** The queue as plain data for a UI: pending titles, per-topic fact counts,
|
|
125
|
-
* skips, and whether the run is complete
|
|
168
|
+
* skips, the two node knobs this run carries, and whether the run is complete
|
|
169
|
+
* (and, if so, whether the node budget is why). Null for no run. */
|
|
126
170
|
export function researchSnapshot(state) {
|
|
127
171
|
if (!state) return null;
|
|
128
172
|
return {
|
|
129
173
|
topic: state.topic,
|
|
130
174
|
limit: state.limit,
|
|
175
|
+
maxDepth: runMaxDepth(state),
|
|
176
|
+
maxTopics: runMaxTopics(state),
|
|
131
177
|
pending: [...state.pending],
|
|
132
178
|
done: state.done.map((d) => ({ title: d.title, facts: d.facts, depth: d.depth })),
|
|
133
179
|
skipped: [...state.skipped],
|
|
134
180
|
complete: state.pending.length === 0,
|
|
181
|
+
nodeCapReached: Boolean(state.nodeCapReached),
|
|
135
182
|
};
|
|
136
183
|
}
|
|
137
184
|
|
|
138
185
|
const totalFacts = (state) => state.done.reduce((sum, d) => sum + d.facts, 0);
|
|
139
186
|
|
|
187
|
+
/** The run's effective knobs, defaulted so a queue resumed from an older
|
|
188
|
+
* persisted file (which carried neither field) reads as today's depth-1
|
|
189
|
+
* behaviour rather than crashing. */
|
|
190
|
+
const runMaxDepth = (state) => (Number.isFinite(state?.maxDepth) ? state.maxDepth : RESEARCH_DEFAULTS.maxDepth);
|
|
191
|
+
const runMaxTopics = (state) => (Number.isFinite(state?.maxTopics) ? state.maxTopics : RESEARCH_DEFAULTS.maxTopics);
|
|
192
|
+
const runFanout = (state) => clampInt(Number.isFinite(state?.limit) ? state.limit : RESEARCH_DEFAULTS.fanoutLimit, 0, RESEARCH_FANOUT_MAX);
|
|
193
|
+
|
|
194
|
+
/** The depth a queued title carries, or 1 for a queue resumed off an older
|
|
195
|
+
* file that never recorded per-title depths. */
|
|
196
|
+
const pendingDepth = (state, title) => {
|
|
197
|
+
const d = state.depths ? state.depths[normFactTerm(title)] : undefined;
|
|
198
|
+
return Number.isFinite(d) ? d : 1;
|
|
199
|
+
};
|
|
200
|
+
|
|
201
|
+
/** Every folded title this run has already touched — the run key, its grounded
|
|
202
|
+
* topics, its skips and its still-pending queue — so a fan-out never re-queues
|
|
203
|
+
* a topic the run has met. */
|
|
204
|
+
function queuedFolds(state) {
|
|
205
|
+
const seen = new Set();
|
|
206
|
+
if (state.key) seen.add(state.key);
|
|
207
|
+
for (const d of state.done) { const f = normFactTerm(d.title); if (f) seen.add(f); }
|
|
208
|
+
for (const t of state.skipped) { const f = normFactTerm(t); if (f) seen.add(f); }
|
|
209
|
+
for (const t of state.pending) { const f = normFactTerm(t); if (f) seen.add(f); }
|
|
210
|
+
return seen;
|
|
211
|
+
}
|
|
212
|
+
|
|
213
|
+
/** Queue `article`'s lead-section links at `fromDepth + 1`, subject to the run's
|
|
214
|
+
* depth ceiling, its per-fan-out cap and — crucially — its TOTAL node budget:
|
|
215
|
+
* the number added never pushes grounded+pending past `maxTopics`. Sets
|
|
216
|
+
* `state.nodeCapReached` when the budget (not the depth, not a lack of links)
|
|
217
|
+
* is what stopped the fan-out, so the progress line can say so. Returns the
|
|
218
|
+
* titles it enqueued. */
|
|
219
|
+
async function enqueueFrom(state, article, fromDepth, provider) {
|
|
220
|
+
const childDepth = fromDepth + 1;
|
|
221
|
+
if (childDepth > runMaxDepth(state)) return [];
|
|
222
|
+
const fanoutCap = runFanout(state);
|
|
223
|
+
if (fanoutCap <= 0 || typeof provider.linkedTitles !== "function") return [];
|
|
224
|
+
const budget = runMaxTopics(state) - (state.done.length + state.pending.length);
|
|
225
|
+
const want = Math.min(fanoutCap, budget);
|
|
226
|
+
if (want <= 0) { state.nodeCapReached = true; return []; }
|
|
227
|
+
let linked = null;
|
|
228
|
+
try { linked = await provider.linkedTitles(article.title, { limit: want + 2 }); } catch { linked = null; }
|
|
229
|
+
const seen = queuedFolds(state);
|
|
230
|
+
if (!state.depths) state.depths = {};
|
|
231
|
+
const added = [];
|
|
232
|
+
for (const title of linked || []) {
|
|
233
|
+
const folded = normFactTerm(title);
|
|
234
|
+
if (!folded || seen.has(folded)) continue;
|
|
235
|
+
seen.add(folded);
|
|
236
|
+
state.pending.push(title);
|
|
237
|
+
state.depths[folded] = childDepth;
|
|
238
|
+
added.push(title);
|
|
239
|
+
if (added.length >= want) break;
|
|
240
|
+
}
|
|
241
|
+
// The budget, not the fan-out cap, was the binding constraint: the run wanted
|
|
242
|
+
// more topics than the node budget would allow and filled to that ceiling.
|
|
243
|
+
if (want < fanoutCap && added.length >= want) state.nodeCapReached = true;
|
|
244
|
+
return added;
|
|
245
|
+
}
|
|
246
|
+
|
|
140
247
|
function progressLine(state) {
|
|
141
|
-
const
|
|
248
|
+
const n = state.done.length;
|
|
249
|
+
const facts = totalFacts(state);
|
|
250
|
+
const done = `${n} topic${n === 1 ? "" : "s"} grounded, ${facts} fact${facts === 1 ? "" : "s"} stored`;
|
|
142
251
|
const skipped = state.skipped.length ? `, ${state.skipped.length} skipped` : "";
|
|
143
|
-
|
|
144
|
-
|
|
252
|
+
const capped = Boolean(state.nodeCapReached);
|
|
253
|
+
if (!state.pending.length) {
|
|
254
|
+
if (capped) return `research on "${state.topic}" reached its node budget — ${done}${skipped}.`;
|
|
255
|
+
return `research on "${state.topic}" is complete — ${done}${skipped}.`;
|
|
256
|
+
}
|
|
257
|
+
const queued = `${state.pending.length} linked topic${state.pending.length === 1 ? "" : "s"} still queued`;
|
|
258
|
+
if (capped) return `${done}${skipped}; ${queued} — "research next" fetches the next one. Node budget of ${runMaxTopics(state)} reached, so no more topics will be added; "research stop" clears the queue.`;
|
|
259
|
+
return `${done}${skipped}; ${queued} — "research next" fetches the next one.`;
|
|
145
260
|
}
|
|
146
261
|
|
|
147
|
-
async function startRun({ topic, limit }, { holder, provider, ingest, config, notify, lexicon }) {
|
|
262
|
+
async function startRun({ topic, limit, depth }, { holder, provider, ingest, config, notify, lexicon }) {
|
|
148
263
|
const key = researchTopicKey(topic, lexicon);
|
|
149
264
|
if (!key) {
|
|
150
265
|
holder.state = null;
|
|
@@ -167,25 +282,17 @@ async function startRun({ topic, limit }, { holder, provider, ingest, config, no
|
|
|
167
282
|
0,
|
|
168
283
|
RESEARCH_FANOUT_MAX,
|
|
169
284
|
);
|
|
170
|
-
|
|
171
|
-
|
|
172
|
-
let linked = null;
|
|
173
|
-
try { linked = await provider.linkedTitles(article.title, { limit: fanout + 2 }); } catch { linked = null; }
|
|
174
|
-
const seen = new Set([key, normFactTerm(article.title)]);
|
|
175
|
-
for (const title of linked || []) {
|
|
176
|
-
const folded = normFactTerm(title);
|
|
177
|
-
if (!folded || seen.has(folded)) continue;
|
|
178
|
-
seen.add(folded);
|
|
179
|
-
pending.push(title);
|
|
180
|
-
if (pending.length >= fanout) break;
|
|
181
|
-
}
|
|
182
|
-
}
|
|
285
|
+
const maxDepth = Number.isFinite(depth) ? clampInt(depth, 0, RESEARCH_MAX_DEPTH) : config.maxDepth;
|
|
286
|
+
const maxTopics = Number.isFinite(config.maxTopics) ? config.maxTopics : RESEARCH_DEFAULTS.maxTopics;
|
|
183
287
|
holder.state = {
|
|
184
|
-
topic, key, title: article.title, limit: fanout,
|
|
185
|
-
pending, done: [{ title: article.title, facts, depth: 0 }],
|
|
288
|
+
topic, key, title: article.title, limit: fanout, maxDepth, maxTopics,
|
|
289
|
+
pending: [], depths: {}, done: [{ title: article.title, facts, depth: 0 }],
|
|
290
|
+
skipped: [], nodeCapReached: false,
|
|
186
291
|
};
|
|
292
|
+
const pending = await enqueueFrom(holder.state, article, 0, provider);
|
|
293
|
+
const depthNote = maxDepth > 1 ? ` following links up to depth ${maxDepth}` : "";
|
|
187
294
|
const queueLine = pending.length
|
|
188
|
-
? `queued ${pending.length} linked topic${pending.length === 1 ? "" : "s"}: ${pending.join(", ")} — "research next" fetches the next one (the page's play button does this for you).`
|
|
295
|
+
? `queued ${pending.length} linked topic${pending.length === 1 ? "" : "s"}: ${pending.join(", ")}${depthNote} — "research next" fetches the next one (the page's play button does this for you).`
|
|
189
296
|
: `no linked topics queued — research on "${topic}" is complete.`;
|
|
190
297
|
return {
|
|
191
298
|
text: `${renderResearchAnswer(key, article)}\nstored ${facts} fact${facts === 1 ? "" : "s"} from "${article.title}". ${queueLine}`,
|
|
@@ -196,7 +303,9 @@ async function startRun({ topic, limit }, { holder, provider, ingest, config, no
|
|
|
196
303
|
async function stepRun({ holder, provider, ingest, notify }) {
|
|
197
304
|
const state = holder.state;
|
|
198
305
|
const title = state.pending[0];
|
|
306
|
+
const depth = pendingDepth(state, title);
|
|
199
307
|
state.pending = state.pending.slice(1);
|
|
308
|
+
if (state.depths) delete state.depths[normFactTerm(title)];
|
|
200
309
|
try { if (typeof notify === "function") notify(title); } catch { /* notify-only */ }
|
|
201
310
|
let article = null;
|
|
202
311
|
try { article = await (provider.pageByTitle ? provider.pageByTitle(title) : provider.lookup(normFactTerm(title))); } catch { article = null; }
|
|
@@ -209,8 +318,9 @@ async function stepRun({ holder, provider, ingest, notify }) {
|
|
|
209
318
|
}
|
|
210
319
|
const key = normFactTerm(article.title) || normFactTerm(title);
|
|
211
320
|
let facts = 0;
|
|
212
|
-
try { facts = await ingest(key, article, researchProvenanceTag(state.key,
|
|
213
|
-
state.done = [...state.done, { title: article.title, facts, depth
|
|
321
|
+
try { facts = await ingest(key, article, researchProvenanceTag(state.key, depth)); } catch { facts = 0; }
|
|
322
|
+
state.done = [...state.done, { title: article.title, facts, depth }];
|
|
323
|
+
if (depth < runMaxDepth(state)) await enqueueFrom(state, article, depth, provider);
|
|
214
324
|
return {
|
|
215
325
|
text: `${renderResearchAnswer(key, article)}\nstored ${facts} fact${facts === 1 ? "" : "s"} from "${article.title}". ${progressLine(state)}`,
|
|
216
326
|
miss: false,
|