dsh-tdai-memory 0.2.13 → 0.3.0

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.
Files changed (72) hide show
  1. package/LICENSE +21 -21
  2. package/README.md +15 -13
  3. package/README.zh.md +9 -7
  4. package/THIRD-PARTY-NOTICES.md +20 -20
  5. package/client.js +21 -4
  6. package/cordis.patch.yml +8 -8
  7. package/index.js +36 -14
  8. package/package.json +77 -78
  9. package/recall-inject.js +86 -86
  10. package/vendor/tdai/adapters/index.js +15 -15
  11. package/vendor/tdai/adapters/openclaw/host-adapter.js +83 -83
  12. package/vendor/tdai/adapters/openclaw/index.js +5 -5
  13. package/vendor/tdai/adapters/openclaw/llm-runner.js +69 -69
  14. package/vendor/tdai/adapters/standalone/host-adapter.js +58 -58
  15. package/vendor/tdai/adapters/standalone/index.js +5 -5
  16. package/vendor/tdai/adapters/standalone/llm-runner.js +34 -6
  17. package/vendor/tdai/config.js +3 -0
  18. package/vendor/tdai/core/conversation/l0-recorder.js +420 -420
  19. package/vendor/tdai/core/hooks/auto-capture.js +236 -236
  20. package/vendor/tdai/core/hooks/auto-recall.js +640 -640
  21. package/vendor/tdai/core/index.js +8 -8
  22. package/vendor/tdai/core/persona/persona-generator.js +184 -184
  23. package/vendor/tdai/core/persona/persona-trigger.js +94 -94
  24. package/vendor/tdai/core/profile/profile-sync.js +200 -200
  25. package/vendor/tdai/core/prompts/l1-dedup.js +72 -72
  26. package/vendor/tdai/core/prompts/l1-extraction.js +34 -34
  27. package/vendor/tdai/core/prompts/persona-generation.js +35 -35
  28. package/vendor/tdai/core/prompts/scene-extraction.js +45 -45
  29. package/vendor/tdai/core/record/l1-dedup.js +318 -318
  30. package/vendor/tdai/core/record/l1-extractor.js +379 -379
  31. package/vendor/tdai/core/record/l1-reader.js +178 -178
  32. package/vendor/tdai/core/record/l1-writer.js +141 -141
  33. package/vendor/tdai/core/report/reporter.js +86 -86
  34. package/vendor/tdai/core/scene/filename-normalizer.js +161 -161
  35. package/vendor/tdai/core/scene/scene-extractor.js +410 -410
  36. package/vendor/tdai/core/scene/scene-format.js +53 -53
  37. package/vendor/tdai/core/scene/scene-index.js +82 -82
  38. package/vendor/tdai/core/scene/scene-navigation.js +57 -57
  39. package/vendor/tdai/core/seed/input.js +404 -404
  40. package/vendor/tdai/core/seed/seed-runtime.js +4 -0
  41. package/vendor/tdai/core/seed/types.js +9 -9
  42. package/vendor/tdai/core/store/bm25-client.js +117 -117
  43. package/vendor/tdai/core/store/bm25-local.js +68 -68
  44. package/vendor/tdai/core/store/embedding.js +632 -632
  45. package/vendor/tdai/core/store/factory.js +95 -95
  46. package/vendor/tdai/core/store/search-utils.js +54 -54
  47. package/vendor/tdai/core/store/sqlite.js +1632 -1632
  48. package/vendor/tdai/core/store/tcvdb-client.js +221 -221
  49. package/vendor/tdai/core/store/tcvdb.js +1047 -1047
  50. package/vendor/tdai/core/store/types.js +17 -17
  51. package/vendor/tdai/core/tdai-core.js +4 -0
  52. package/vendor/tdai/core/tools/conversation-search.js +199 -199
  53. package/vendor/tdai/core/tools/memory-search.js +207 -207
  54. package/vendor/tdai/core/types.js +12 -12
  55. package/vendor/tdai/utils/backup.js +189 -189
  56. package/vendor/tdai/utils/checkpoint.js +360 -360
  57. package/vendor/tdai/utils/clean-context-runner.js +472 -472
  58. package/vendor/tdai/utils/ensure-hook-policy.js +213 -213
  59. package/vendor/tdai/utils/env.js +13 -13
  60. package/vendor/tdai/utils/managed-timer.js +127 -127
  61. package/vendor/tdai/utils/manifest.js +91 -91
  62. package/vendor/tdai/utils/memory-cleaner.js +300 -300
  63. package/vendor/tdai/utils/no-think-fetch.js +109 -109
  64. package/vendor/tdai/utils/openclaw-state-dir.js +26 -26
  65. package/vendor/tdai/utils/pipeline-factory.js +517 -517
  66. package/vendor/tdai/utils/pipeline-manager.js +892 -892
  67. package/vendor/tdai/utils/sanitize.js +360 -360
  68. package/vendor/tdai/utils/serial-queue.js +103 -103
  69. package/vendor/tdai/utils/session-filter.js +88 -88
  70. package/vendor/tdai/utils/session-header-fetch.js +88 -0
  71. package/vendor/tdai/utils/text-utils.js +29 -29
  72. package/vendor/tdai/utils/time.js +252 -252
@@ -1,103 +1,103 @@
1
- /**
2
- * SerialQueue: a lightweight task queue with concurrency=1.
3
- *
4
- * Equivalent to `new PQueue({ concurrency: 1 })` but with zero external
5
- * dependencies. Supports:
6
- * - Serial execution (FIFO)
7
- * - `add(fn)` to enqueue a task (returns the task's result promise)
8
- * - `onIdle()` to wait until all queued tasks have completed
9
- * - `pause()` / `start()` to suspend/resume execution
10
- * - `size` to check pending task count
11
- * - Optional debug logger for enqueue/dequeue/complete diagnostics
12
- */
13
- export class SerialQueue {
14
- /** Human-readable name for logging / diagnostics. */
15
- name;
16
- queue = [];
17
- running = false;
18
- paused = false;
19
- idleResolvers = [];
20
- /** Optional debug logger — receives diagnostic messages for enqueue/dequeue/complete. */
21
- debugFn;
22
- constructor(name = "unnamed") {
23
- this.name = name;
24
- }
25
- /** Set a debug logger for queue diagnostics. */
26
- setDebugLogger(fn) {
27
- this.debugFn = fn;
28
- }
29
- /** Number of tasks waiting to be executed. */
30
- get size() {
31
- return this.queue.length;
32
- }
33
- /** Whether a task is currently executing. */
34
- get pending() {
35
- return this.running;
36
- }
37
- /** Whether the queue is idle (no queued tasks and nothing running). */
38
- get idle() {
39
- return this.queue.length === 0 && !this.running;
40
- }
41
- /** Add a task to the queue. Returns the task's result promise. */
42
- add(task) {
43
- return new Promise((resolve, reject) => {
44
- this.queue.push({
45
- task: task,
46
- resolve: resolve,
47
- reject,
48
- });
49
- this.debugFn?.(`[queue:${this.name}] enqueued, pending=${this.queue.length}, running=${this.running}`);
50
- this.drain();
51
- });
52
- }
53
- /** Pause the queue. Currently running task will finish, but no new tasks start. */
54
- pause() {
55
- this.paused = true;
56
- }
57
- /** Resume the queue after pause(). */
58
- start() {
59
- this.paused = false;
60
- this.drain();
61
- }
62
- /** Returns a promise that resolves when all queued tasks have completed. */
63
- onIdle() {
64
- if (this.queue.length === 0 && !this.running) {
65
- return Promise.resolve();
66
- }
67
- return new Promise((resolve) => {
68
- this.idleResolvers.push(resolve);
69
- });
70
- }
71
- /** Clear all pending (not yet started) tasks. */
72
- clear() {
73
- for (const entry of this.queue) {
74
- entry.reject(new Error("Queue cleared"));
75
- }
76
- this.queue = [];
77
- }
78
- drain() {
79
- if (this.running || this.paused || this.queue.length === 0)
80
- return;
81
- const entry = this.queue.shift();
82
- this.running = true;
83
- this.debugFn?.(`[queue:${this.name}] dequeued, starting execution (remaining=${this.queue.length})`);
84
- entry
85
- .task()
86
- .then((result) => entry.resolve(result))
87
- .catch((err) => entry.reject(err))
88
- .finally(() => {
89
- this.running = false;
90
- this.debugFn?.(`[queue:${this.name}] task completed (remaining=${this.queue.length})`);
91
- if (this.queue.length === 0) {
92
- // Notify idle waiters
93
- const resolvers = this.idleResolvers;
94
- this.idleResolvers = [];
95
- for (const resolve of resolvers)
96
- resolve();
97
- }
98
- else {
99
- this.drain();
100
- }
101
- });
102
- }
103
- }
1
+ /**
2
+ * SerialQueue: a lightweight task queue with concurrency=1.
3
+ *
4
+ * Equivalent to `new PQueue({ concurrency: 1 })` but with zero external
5
+ * dependencies. Supports:
6
+ * - Serial execution (FIFO)
7
+ * - `add(fn)` to enqueue a task (returns the task's result promise)
8
+ * - `onIdle()` to wait until all queued tasks have completed
9
+ * - `pause()` / `start()` to suspend/resume execution
10
+ * - `size` to check pending task count
11
+ * - Optional debug logger for enqueue/dequeue/complete diagnostics
12
+ */
13
+ export class SerialQueue {
14
+ /** Human-readable name for logging / diagnostics. */
15
+ name;
16
+ queue = [];
17
+ running = false;
18
+ paused = false;
19
+ idleResolvers = [];
20
+ /** Optional debug logger — receives diagnostic messages for enqueue/dequeue/complete. */
21
+ debugFn;
22
+ constructor(name = "unnamed") {
23
+ this.name = name;
24
+ }
25
+ /** Set a debug logger for queue diagnostics. */
26
+ setDebugLogger(fn) {
27
+ this.debugFn = fn;
28
+ }
29
+ /** Number of tasks waiting to be executed. */
30
+ get size() {
31
+ return this.queue.length;
32
+ }
33
+ /** Whether a task is currently executing. */
34
+ get pending() {
35
+ return this.running;
36
+ }
37
+ /** Whether the queue is idle (no queued tasks and nothing running). */
38
+ get idle() {
39
+ return this.queue.length === 0 && !this.running;
40
+ }
41
+ /** Add a task to the queue. Returns the task's result promise. */
42
+ add(task) {
43
+ return new Promise((resolve, reject) => {
44
+ this.queue.push({
45
+ task: task,
46
+ resolve: resolve,
47
+ reject,
48
+ });
49
+ this.debugFn?.(`[queue:${this.name}] enqueued, pending=${this.queue.length}, running=${this.running}`);
50
+ this.drain();
51
+ });
52
+ }
53
+ /** Pause the queue. Currently running task will finish, but no new tasks start. */
54
+ pause() {
55
+ this.paused = true;
56
+ }
57
+ /** Resume the queue after pause(). */
58
+ start() {
59
+ this.paused = false;
60
+ this.drain();
61
+ }
62
+ /** Returns a promise that resolves when all queued tasks have completed. */
63
+ onIdle() {
64
+ if (this.queue.length === 0 && !this.running) {
65
+ return Promise.resolve();
66
+ }
67
+ return new Promise((resolve) => {
68
+ this.idleResolvers.push(resolve);
69
+ });
70
+ }
71
+ /** Clear all pending (not yet started) tasks. */
72
+ clear() {
73
+ for (const entry of this.queue) {
74
+ entry.reject(new Error("Queue cleared"));
75
+ }
76
+ this.queue = [];
77
+ }
78
+ drain() {
79
+ if (this.running || this.paused || this.queue.length === 0)
80
+ return;
81
+ const entry = this.queue.shift();
82
+ this.running = true;
83
+ this.debugFn?.(`[queue:${this.name}] dequeued, starting execution (remaining=${this.queue.length})`);
84
+ entry
85
+ .task()
86
+ .then((result) => entry.resolve(result))
87
+ .catch((err) => entry.reject(err))
88
+ .finally(() => {
89
+ this.running = false;
90
+ this.debugFn?.(`[queue:${this.name}] task completed (remaining=${this.queue.length})`);
91
+ if (this.queue.length === 0) {
92
+ // Notify idle waiters
93
+ const resolvers = this.idleResolvers;
94
+ this.idleResolvers = [];
95
+ for (const resolve of resolvers)
96
+ resolve();
97
+ }
98
+ else {
99
+ this.drain();
100
+ }
101
+ });
102
+ }
103
+ }
@@ -1,88 +1,88 @@
1
- /**
2
- * Session filtering for memory-tdai.
3
- *
4
- * Decides whether a session should be ignored by the memory plugin
5
- * (capture, recall, pipeline scheduling). All skip rules are compiled
6
- * into a flat list of matchers at construction time — zero per-call overhead.
7
- */
8
- // ============================
9
- // Non-interactive trigger detection
10
- // ============================
11
- const SKIP_TRIGGERS = new Set(["cron", "heartbeat", "automation", "schedule"]);
12
- /**
13
- * Returns true when the hook was fired by a non-interactive trigger
14
- * (heartbeat, cron job, automation, etc.) — these produce no meaningful
15
- * user conversation and should not be captured or counted.
16
- */
17
- export function isNonInteractiveTrigger(trigger, sessionKey) {
18
- if (trigger && SKIP_TRIGGERS.has(trigger.toLowerCase()))
19
- return true;
20
- if (sessionKey) {
21
- if (/:cron:/i.test(sessionKey) || /:heartbeat:/i.test(sessionKey))
22
- return true;
23
- }
24
- return false;
25
- }
26
- // ============================
27
- // Built-in skip rules (always active)
28
- // ============================
29
- /**
30
- * Hard-coded matchers that identify internal / non-user sessions.
31
- * These are always applied regardless of user configuration.
32
- */
33
- const BUILTIN_MATCHERS = [
34
- // Scene extraction runner sessions
35
- (key) => key.includes(":memory-scene-extract-"),
36
- // OpenClaw subagent sessions
37
- (key) => key.includes(":subagent:"),
38
- // Temporary / internal utility sessions (e.g. temp:slug-generator)
39
- (key) => key.startsWith("temp:"),
40
- ];
41
- // ============================
42
- // Glob → matcher compiler
43
- // ============================
44
- /**
45
- * Turn a simple glob pattern (only `*` supported) into a matcher
46
- * that tests the full sessionKey.
47
- *
48
- * Since sessionKeys look like `agent:<agentId>:...`, we match the
49
- * glob against the whole key so users can write patterns like
50
- * `bench-judge-*` (matched anywhere) or more specific ones.
51
- */
52
- function globToMatcher(pattern) {
53
- const escaped = pattern.replace(/[.+^${}()|[\]\\]/g, "\\$&").replace(/\*/g, ".*");
54
- const re = new RegExp(escaped);
55
- return (key) => re.test(key);
56
- }
57
- // ============================
58
- // SessionFilter
59
- // ============================
60
- /**
61
- * Unified filter: construct once at plugin startup, then call
62
- * `shouldSkip(sessionKey)` or `shouldSkipCtx(ctx)` at each gate.
63
- */
64
- export class SessionFilter {
65
- matchers;
66
- constructor(excludeAgents = []) {
67
- // Merge built-in rules + user-configured exclude patterns into one flat list
68
- const userMatchers = excludeAgents
69
- .map((p) => p.trim())
70
- .filter((p) => p.length > 0)
71
- .map(globToMatcher);
72
- this.matchers = [...BUILTIN_MATCHERS, ...userMatchers];
73
- }
74
- /** Should this sessionKey be skipped? */
75
- shouldSkip(sessionKey) {
76
- return this.matchers.some((m) => m(sessionKey));
77
- }
78
- /** Should this hook context be skipped? */
79
- shouldSkipCtx(ctx) {
80
- if (!ctx.sessionKey)
81
- return true;
82
- if (ctx.sessionId?.startsWith("memory-"))
83
- return true;
84
- if (isNonInteractiveTrigger(ctx.trigger, ctx.sessionKey))
85
- return true;
86
- return this.shouldSkip(ctx.sessionKey);
87
- }
88
- }
1
+ /**
2
+ * Session filtering for memory-tdai.
3
+ *
4
+ * Decides whether a session should be ignored by the memory plugin
5
+ * (capture, recall, pipeline scheduling). All skip rules are compiled
6
+ * into a flat list of matchers at construction time — zero per-call overhead.
7
+ */
8
+ // ============================
9
+ // Non-interactive trigger detection
10
+ // ============================
11
+ const SKIP_TRIGGERS = new Set(["cron", "heartbeat", "automation", "schedule"]);
12
+ /**
13
+ * Returns true when the hook was fired by a non-interactive trigger
14
+ * (heartbeat, cron job, automation, etc.) — these produce no meaningful
15
+ * user conversation and should not be captured or counted.
16
+ */
17
+ export function isNonInteractiveTrigger(trigger, sessionKey) {
18
+ if (trigger && SKIP_TRIGGERS.has(trigger.toLowerCase()))
19
+ return true;
20
+ if (sessionKey) {
21
+ if (/:cron:/i.test(sessionKey) || /:heartbeat:/i.test(sessionKey))
22
+ return true;
23
+ }
24
+ return false;
25
+ }
26
+ // ============================
27
+ // Built-in skip rules (always active)
28
+ // ============================
29
+ /**
30
+ * Hard-coded matchers that identify internal / non-user sessions.
31
+ * These are always applied regardless of user configuration.
32
+ */
33
+ const BUILTIN_MATCHERS = [
34
+ // Scene extraction runner sessions
35
+ (key) => key.includes(":memory-scene-extract-"),
36
+ // OpenClaw subagent sessions
37
+ (key) => key.includes(":subagent:"),
38
+ // Temporary / internal utility sessions (e.g. temp:slug-generator)
39
+ (key) => key.startsWith("temp:"),
40
+ ];
41
+ // ============================
42
+ // Glob → matcher compiler
43
+ // ============================
44
+ /**
45
+ * Turn a simple glob pattern (only `*` supported) into a matcher
46
+ * that tests the full sessionKey.
47
+ *
48
+ * Since sessionKeys look like `agent:<agentId>:...`, we match the
49
+ * glob against the whole key so users can write patterns like
50
+ * `bench-judge-*` (matched anywhere) or more specific ones.
51
+ */
52
+ function globToMatcher(pattern) {
53
+ const escaped = pattern.replace(/[.+^${}()|[\]\\]/g, "\\$&").replace(/\*/g, ".*");
54
+ const re = new RegExp(escaped);
55
+ return (key) => re.test(key);
56
+ }
57
+ // ============================
58
+ // SessionFilter
59
+ // ============================
60
+ /**
61
+ * Unified filter: construct once at plugin startup, then call
62
+ * `shouldSkip(sessionKey)` or `shouldSkipCtx(ctx)` at each gate.
63
+ */
64
+ export class SessionFilter {
65
+ matchers;
66
+ constructor(excludeAgents = []) {
67
+ // Merge built-in rules + user-configured exclude patterns into one flat list
68
+ const userMatchers = excludeAgents
69
+ .map((p) => p.trim())
70
+ .filter((p) => p.length > 0)
71
+ .map(globToMatcher);
72
+ this.matchers = [...BUILTIN_MATCHERS, ...userMatchers];
73
+ }
74
+ /** Should this sessionKey be skipped? */
75
+ shouldSkip(sessionKey) {
76
+ return this.matchers.some((m) => m(sessionKey));
77
+ }
78
+ /** Should this hook context be skipped? */
79
+ shouldSkipCtx(ctx) {
80
+ if (!ctx.sessionKey)
81
+ return true;
82
+ if (ctx.sessionId?.startsWith("memory-"))
83
+ return true;
84
+ if (isNonInteractiveTrigger(ctx.trigger, ctx.sessionKey))
85
+ return true;
86
+ return this.shouldSkip(ctx.sessionKey);
87
+ }
88
+ }
@@ -0,0 +1,88 @@
1
+ /**
2
+ * Fetch wrapper that injects a stable session-id header (`x-opencode-session`)
3
+ * into outgoing LLM requests.
4
+ *
5
+ * OpenCode Go (and other OpenAI-compatible gateways) require every request to
6
+ * carry a stable per-conversation id; requests without it started erroring
7
+ * after 2026-09-06. The value is:
8
+ *
9
+ * 1. An explicit configured `sessionId`, when set;
10
+ * 2. Otherwise a persistent per-instance random id. The id is generated once
11
+ * and stored under the TDAI data dir (`<dataDir>/session-id`), so it
12
+ * survives restarts — one stable id per plugin instance, which is exactly
13
+ * what the gateway asks for when the client has no real conversation.
14
+ *
15
+ * Embedding and other non-chat traffic still gets the header (harmless, and
16
+ * gateways may require it on every request). When disabled, returns
17
+ * `globalThis.fetch` unchanged.
18
+ */
19
+ import { randomUUID } from "node:crypto";
20
+ import { join } from "node:path";
21
+ import { mkdir, readFile, writeFile } from "node:fs/promises";
22
+
23
+ const SESSION_ID_FILE = "session-id";
24
+
25
+ let cachedId = "";
26
+ /** Load-or-create a persistent per-instance session id under `dataDir`. */
27
+ export async function resolvePersistentSessionId(dataDir) {
28
+ if (cachedId)
29
+ return cachedId;
30
+ if (!dataDir) {
31
+ cachedId = `tdai-memory-${randomUUID()}`;
32
+ return cachedId;
33
+ }
34
+ try {
35
+ const file = join(dataDir, SESSION_ID_FILE);
36
+ const existing = await readFile(file, "utf8").catch(() => "");
37
+ const trimmed = existing.trim();
38
+ if (trimmed) {
39
+ cachedId = trimmed;
40
+ return cachedId;
41
+ }
42
+ const fresh = `tdai-memory-${randomUUID()}`;
43
+ await mkdir(dataDir, { recursive: true });
44
+ await writeFile(file, fresh, "utf8");
45
+ cachedId = fresh;
46
+ return cachedId;
47
+ }
48
+ catch {
49
+ // Any persistence failure degrades to a per-process id.
50
+ cachedId = `tdai-memory-${randomUUID()}`;
51
+ return cachedId;
52
+ }
53
+ }
54
+
55
+ /**
56
+ * Build a session-id header value for a request.
57
+ * @param {object} opts { send, headerName, sessionId, persistentId }
58
+ * @returns {Record<string,string>|undefined} header object, or undefined when disabled.
59
+ */
60
+ export function sessionHeaderOf({ send, headerName, sessionId, persistentId }) {
61
+ if (send === false)
62
+ return undefined;
63
+ const id = (sessionId && String(sessionId).trim()) || persistentId || "";
64
+ if (!id)
65
+ return undefined;
66
+ return { [headerName || "x-opencode-session"]: id };
67
+ }
68
+
69
+ /**
70
+ * Create a fetch wrapper that injects the session-id header into every request.
71
+ * @param {object} opts { send, headerName, sessionId, persistentId }
72
+ * @returns {typeof globalThis.fetch}
73
+ */
74
+ export function createSessionHeaderFetch({ send, headerName, sessionId, persistentId }) {
75
+ if (send === false)
76
+ return globalThis.fetch;
77
+ const header = sessionHeaderOf({ send, headerName, sessionId, persistentId });
78
+ if (!header)
79
+ return globalThis.fetch;
80
+ return (async (input, init) => {
81
+ const next = init ? { ...init } : {};
82
+ next.headers = new Headers(init?.headers);
83
+ for (const [k, v] of Object.entries(header)) {
84
+ next.headers.set(k, v);
85
+ }
86
+ return globalThis.fetch(input, next);
87
+ });
88
+ }
@@ -1,29 +1,29 @@
1
- /**
2
- * Shared text utility functions for the memory-tdai plugin.
3
- */
4
- /**
5
- * Extract meaningful words from text (supports CJK and Latin).
6
- *
7
- * Used by both auto-recall (keyword search) and l1-dedup (keyword candidate recall).
8
- * Extracted to a shared module to prevent implementation drift.
9
- */
10
- export function extractWords(text) {
11
- const words = new Set();
12
- // Latin words (2+ chars)
13
- const latinWords = text.toLowerCase().match(/[a-z0-9]{2,}/g);
14
- if (latinWords) {
15
- for (const w of latinWords)
16
- words.add(w);
17
- }
18
- // CJK characters (each char as a "word", plus 2-gram)
19
- const cjkChars = text.match(/[\u4e00-\u9fff\u3040-\u30ff\uac00-\ud7af]/g);
20
- if (cjkChars) {
21
- for (const c of cjkChars)
22
- words.add(c);
23
- // 2-grams for better matching
24
- for (let i = 0; i < cjkChars.length - 1; i++) {
25
- words.add(cjkChars[i] + cjkChars[i + 1]);
26
- }
27
- }
28
- return words;
29
- }
1
+ /**
2
+ * Shared text utility functions for the memory-tdai plugin.
3
+ */
4
+ /**
5
+ * Extract meaningful words from text (supports CJK and Latin).
6
+ *
7
+ * Used by both auto-recall (keyword search) and l1-dedup (keyword candidate recall).
8
+ * Extracted to a shared module to prevent implementation drift.
9
+ */
10
+ export function extractWords(text) {
11
+ const words = new Set();
12
+ // Latin words (2+ chars)
13
+ const latinWords = text.toLowerCase().match(/[a-z0-9]{2,}/g);
14
+ if (latinWords) {
15
+ for (const w of latinWords)
16
+ words.add(w);
17
+ }
18
+ // CJK characters (each char as a "word", plus 2-gram)
19
+ const cjkChars = text.match(/[\u4e00-\u9fff\u3040-\u30ff\uac00-\ud7af]/g);
20
+ if (cjkChars) {
21
+ for (const c of cjkChars)
22
+ words.add(c);
23
+ // 2-grams for better matching
24
+ for (let i = 0; i < cjkChars.length - 1; i++) {
25
+ words.add(cjkChars[i] + cjkChars[i + 1]);
26
+ }
27
+ }
28
+ return words;
29
+ }