clawvault 3.1.0 → 3.2.1

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 (273) hide show
  1. package/README.md +422 -141
  2. package/bin/clawvault.js +10 -2
  3. package/bin/command-registration.test.js +3 -1
  4. package/bin/command-runtime.js +9 -1
  5. package/bin/register-core-commands.js +23 -28
  6. package/bin/register-maintenance-commands.js +39 -3
  7. package/bin/register-query-commands.js +58 -29
  8. package/bin/register-tailscale-commands.js +106 -0
  9. package/bin/register-task-commands.js +18 -1
  10. package/bin/register-task-commands.test.js +16 -0
  11. package/bin/register-vault-operations-commands.js +29 -1
  12. package/bin/register-workgraph-commands.js +451 -0
  13. package/dashboard/lib/graph-diff.js +104 -0
  14. package/dashboard/lib/graph-diff.test.js +75 -0
  15. package/dashboard/lib/vault-parser.js +556 -0
  16. package/dashboard/lib/vault-parser.test.js +254 -0
  17. package/dashboard/public/app.js +796 -0
  18. package/dashboard/public/index.html +52 -0
  19. package/dashboard/public/styles.css +221 -0
  20. package/dashboard/server.js +374 -0
  21. package/dist/{chunk-C7OK5WKP.js → chunk-2JQ3O2YL.js} +4 -4
  22. package/dist/{chunk-VR5NE7PZ.js → chunk-2RAZ4ZFE.js} +1 -1
  23. package/dist/{chunk-F2JEUD4J.js → chunk-4ITRXIVT.js} +5 -7
  24. package/dist/{chunk-GUKMRGM7.js → chunk-4OXMU5S2.js} +1 -1
  25. package/dist/chunk-5PJ4STIC.js +465 -0
  26. package/dist/{chunk-62YTUT6J.js → chunk-AZYOKJYC.js} +2 -2
  27. package/dist/chunk-BSJ6RIT7.js +447 -0
  28. package/dist/chunk-ECRZL5XR.js +50 -0
  29. package/dist/chunk-ERNE2FZ5.js +189 -0
  30. package/dist/{chunk-WAZ3NLWL.js → chunk-F55HGNU4.js} +0 -47
  31. package/dist/{chunk-VGLOTGAS.js → chunk-FAKNOB7Y.js} +2 -2
  32. package/dist/{chunk-QK3UCXWL.js → chunk-FHFUXL6G.js} +2 -2
  33. package/dist/chunk-GNJL4YGR.js +79 -0
  34. package/dist/chunk-HR4KN6S2.js +152 -0
  35. package/dist/{chunk-OZ7RIXTO.js → chunk-IIOU45CK.js} +1 -1
  36. package/dist/chunk-IJBFGPCS.js +33 -0
  37. package/dist/chunk-IVRIKYFE.js +520 -0
  38. package/dist/chunk-K7PNYS45.js +93 -0
  39. package/dist/chunk-MDIH26GC.js +183 -0
  40. package/dist/{chunk-LYHGEHXG.js → chunk-MFAWT5O5.js} +0 -1
  41. package/dist/{chunk-H34S76MB.js → chunk-MNPUYCHQ.js} +6 -6
  42. package/dist/chunk-NTOPJI7W.js +207 -0
  43. package/dist/{chunk-QBLMXKF2.js → chunk-OIWVQYQF.js} +1 -1
  44. package/dist/chunk-PG56HX5T.js +154 -0
  45. package/dist/{chunk-LNJA2UGL.js → chunk-PI4WMLMG.js} +7 -84
  46. package/dist/chunk-QMHPQYUV.js +363 -0
  47. package/dist/{chunk-H62BP7RI.js → chunk-QPDDIHXE.js} +209 -43
  48. package/dist/{chunk-N2AXRYLC.js → chunk-QWQ3TIKS.js} +1 -1
  49. package/dist/{chunk-3DHXQHYG.js → chunk-R2MIW5G7.js} +1 -1
  50. package/dist/{chunk-SJSFRIYS.js → chunk-S5OJEGFG.js} +2 -2
  51. package/dist/chunk-SS4B7P7V.js +99 -0
  52. package/dist/chunk-TIGW564L.js +628 -0
  53. package/dist/chunk-U67V476Y.js +35 -0
  54. package/dist/{chunk-JY6FYXIT.js → chunk-UCQAOZHW.js} +6 -11
  55. package/dist/{chunk-ITPEXLHA.js → chunk-URXDAUVH.js} +24 -5
  56. package/dist/chunk-WIOLLGAD.js +190 -0
  57. package/dist/{chunk-3WRJEKN4.js → chunk-WJVWINEM.js} +72 -8
  58. package/dist/chunk-WMGIIABP.js +15 -0
  59. package/dist/{chunk-33UGEQRT.js → chunk-X3SPPUFG.js} +151 -64
  60. package/dist/{chunk-3NSBOUT3.js → chunk-Y3TIJEBP.js} +314 -79
  61. package/dist/chunk-Y6VJKXGL.js +373 -0
  62. package/dist/{chunk-LI4O6NVK.js → chunk-YDWHS4LJ.js} +49 -9
  63. package/dist/{chunk-U55BGUAU.js → chunk-YNIPYN4F.js} +5 -5
  64. package/dist/chunk-YXQCA6B7.js +226 -0
  65. package/dist/cli/index.js +26 -22
  66. package/dist/commands/archive.js +3 -3
  67. package/dist/commands/backlog.js +3 -3
  68. package/dist/commands/blocked.js +3 -3
  69. package/dist/commands/canvas.d.ts +15 -0
  70. package/dist/commands/canvas.js +200 -0
  71. package/dist/commands/checkpoint.js +2 -2
  72. package/dist/commands/compat.js +2 -2
  73. package/dist/commands/context.js +7 -5
  74. package/dist/commands/doctor.d.ts +11 -7
  75. package/dist/commands/doctor.js +16 -14
  76. package/dist/commands/embed.js +5 -6
  77. package/dist/commands/entities.js +2 -2
  78. package/dist/commands/graph.js +3 -3
  79. package/dist/commands/inject.d.ts +1 -1
  80. package/dist/commands/inject.js +4 -5
  81. package/dist/commands/kanban.js +4 -4
  82. package/dist/commands/link.js +2 -2
  83. package/dist/commands/migrate-observations.js +4 -4
  84. package/dist/commands/observe.d.ts +0 -1
  85. package/dist/commands/observe.js +13 -12
  86. package/dist/commands/project.js +5 -5
  87. package/dist/commands/rebuild-embeddings.d.ts +21 -0
  88. package/dist/commands/rebuild-embeddings.js +91 -0
  89. package/dist/commands/rebuild.js +12 -11
  90. package/dist/commands/recover.js +3 -3
  91. package/dist/commands/reflect.js +6 -7
  92. package/dist/commands/repair-session.js +1 -1
  93. package/dist/commands/replay.js +14 -14
  94. package/dist/commands/session-recap.js +1 -1
  95. package/dist/commands/setup.d.ts +2 -89
  96. package/dist/commands/setup.js +3 -21
  97. package/dist/commands/shell-init.js +1 -1
  98. package/dist/commands/sleep.d.ts +1 -1
  99. package/dist/commands/sleep.js +18 -17
  100. package/dist/commands/status.d.ts +2 -0
  101. package/dist/commands/status.js +40 -30
  102. package/dist/commands/sync-bd.d.ts +10 -0
  103. package/dist/commands/sync-bd.js +10 -0
  104. package/dist/commands/tailscale.d.ts +52 -0
  105. package/dist/commands/tailscale.js +26 -0
  106. package/dist/commands/task.js +4 -4
  107. package/dist/commands/template.js +2 -2
  108. package/dist/commands/wake.d.ts +1 -1
  109. package/dist/commands/wake.js +11 -10
  110. package/dist/index.d.ts +334 -191
  111. package/dist/index.js +432 -108
  112. package/dist/{inject-Bzi5E-By.d.ts → inject-DYUrDqQO.d.ts} +3 -3
  113. package/dist/ledger-B7g7jhqG.d.ts +44 -0
  114. package/dist/lib/auto-linker.js +1 -1
  115. package/dist/lib/canvas-layout.d.ts +115 -0
  116. package/dist/lib/canvas-layout.js +35 -0
  117. package/dist/lib/config.d.ts +27 -3
  118. package/dist/lib/config.js +4 -2
  119. package/dist/lib/entity-index.js +1 -1
  120. package/dist/lib/project-utils.js +4 -4
  121. package/dist/lib/session-repair.js +1 -1
  122. package/dist/lib/session-utils.js +1 -1
  123. package/dist/lib/tailscale.d.ts +225 -0
  124. package/dist/lib/tailscale.js +50 -0
  125. package/dist/lib/task-utils.js +3 -3
  126. package/dist/lib/template-engine.js +1 -1
  127. package/dist/lib/webdav.d.ts +109 -0
  128. package/dist/lib/webdav.js +35 -0
  129. package/dist/plugin/index.d.ts +344 -28
  130. package/dist/plugin/index.js +3919 -227
  131. package/dist/registry-BR4326o0.d.ts +30 -0
  132. package/dist/store-CA-6sKCJ.d.ts +34 -0
  133. package/dist/thread-B9LhXNU0.d.ts +41 -0
  134. package/dist/{types-Y2_Um2Ls.d.ts → types-BbWJoC1c.d.ts} +1 -44
  135. package/dist/workgraph/index.d.ts +5 -0
  136. package/dist/workgraph/index.js +23 -0
  137. package/dist/workgraph/ledger.d.ts +2 -0
  138. package/dist/workgraph/ledger.js +25 -0
  139. package/dist/workgraph/registry.d.ts +2 -0
  140. package/dist/workgraph/registry.js +19 -0
  141. package/dist/workgraph/store.d.ts +2 -0
  142. package/dist/workgraph/store.js +25 -0
  143. package/dist/workgraph/thread.d.ts +2 -0
  144. package/dist/workgraph/thread.js +25 -0
  145. package/dist/workgraph/types.d.ts +54 -0
  146. package/dist/workgraph/types.js +7 -0
  147. package/hooks/clawvault/HOOK.md +113 -0
  148. package/hooks/clawvault/handler.js +1559 -0
  149. package/hooks/clawvault/handler.test.js +510 -0
  150. package/hooks/clawvault/openclaw.plugin.json +72 -0
  151. package/openclaw.plugin.json +235 -30
  152. package/package.json +20 -20
  153. package/dist/chunk-3RG5ZIWI.js +0 -10
  154. package/dist/chunk-3ZIH425O.js +0 -871
  155. package/dist/chunk-6U6MK36V.js +0 -205
  156. package/dist/chunk-CMB7UL7C.js +0 -327
  157. package/dist/chunk-D2H45LON.js +0 -1074
  158. package/dist/chunk-E7MFQB6D.js +0 -163
  159. package/dist/chunk-GQSLDZTS.js +0 -560
  160. package/dist/chunk-MFM6K7PU.js +0 -374
  161. package/dist/chunk-MXSSG3QU.js +0 -42
  162. package/dist/chunk-OCGVIN3L.js +0 -88
  163. package/dist/chunk-PAH27GSN.js +0 -108
  164. package/dist/chunk-YCUNCH2I.js +0 -78
  165. package/dist/cli/index.cjs +0 -8584
  166. package/dist/cli/index.d.cts +0 -5
  167. package/dist/commands/archive.cjs +0 -287
  168. package/dist/commands/archive.d.cts +0 -11
  169. package/dist/commands/backlog.cjs +0 -721
  170. package/dist/commands/backlog.d.cts +0 -53
  171. package/dist/commands/blocked.cjs +0 -204
  172. package/dist/commands/blocked.d.cts +0 -26
  173. package/dist/commands/checkpoint.cjs +0 -244
  174. package/dist/commands/checkpoint.d.cts +0 -41
  175. package/dist/commands/compat.cjs +0 -294
  176. package/dist/commands/compat.d.cts +0 -28
  177. package/dist/commands/context.cjs +0 -2990
  178. package/dist/commands/context.d.cts +0 -2
  179. package/dist/commands/doctor.cjs +0 -2986
  180. package/dist/commands/doctor.d.cts +0 -21
  181. package/dist/commands/embed.cjs +0 -232
  182. package/dist/commands/embed.d.cts +0 -17
  183. package/dist/commands/entities.cjs +0 -141
  184. package/dist/commands/entities.d.cts +0 -7
  185. package/dist/commands/graph.cjs +0 -501
  186. package/dist/commands/graph.d.cts +0 -21
  187. package/dist/commands/inject.cjs +0 -1636
  188. package/dist/commands/inject.d.cts +0 -2
  189. package/dist/commands/kanban.cjs +0 -884
  190. package/dist/commands/kanban.d.cts +0 -63
  191. package/dist/commands/link.cjs +0 -965
  192. package/dist/commands/link.d.cts +0 -11
  193. package/dist/commands/migrate-observations.cjs +0 -362
  194. package/dist/commands/migrate-observations.d.cts +0 -19
  195. package/dist/commands/observe.cjs +0 -4099
  196. package/dist/commands/observe.d.cts +0 -23
  197. package/dist/commands/project.cjs +0 -1341
  198. package/dist/commands/project.d.cts +0 -85
  199. package/dist/commands/rebuild.cjs +0 -3136
  200. package/dist/commands/rebuild.d.cts +0 -11
  201. package/dist/commands/recover.cjs +0 -361
  202. package/dist/commands/recover.d.cts +0 -38
  203. package/dist/commands/reflect.cjs +0 -1008
  204. package/dist/commands/reflect.d.cts +0 -11
  205. package/dist/commands/repair-session.cjs +0 -457
  206. package/dist/commands/repair-session.d.cts +0 -38
  207. package/dist/commands/replay.cjs +0 -4103
  208. package/dist/commands/replay.d.cts +0 -16
  209. package/dist/commands/session-recap.cjs +0 -353
  210. package/dist/commands/session-recap.d.cts +0 -27
  211. package/dist/commands/setup.cjs +0 -1278
  212. package/dist/commands/setup.d.cts +0 -99
  213. package/dist/commands/shell-init.cjs +0 -75
  214. package/dist/commands/shell-init.d.cts +0 -7
  215. package/dist/commands/sleep.cjs +0 -6029
  216. package/dist/commands/sleep.d.cts +0 -36
  217. package/dist/commands/status.cjs +0 -2737
  218. package/dist/commands/status.d.cts +0 -52
  219. package/dist/commands/task.cjs +0 -1236
  220. package/dist/commands/task.d.cts +0 -97
  221. package/dist/commands/template.cjs +0 -457
  222. package/dist/commands/template.d.cts +0 -36
  223. package/dist/commands/wake.cjs +0 -2627
  224. package/dist/commands/wake.d.cts +0 -22
  225. package/dist/context-BUGaWpyL.d.cts +0 -46
  226. package/dist/index.cjs +0 -12373
  227. package/dist/index.d.cts +0 -854
  228. package/dist/inject-Bzi5E-By.d.cts +0 -137
  229. package/dist/lib/auto-linker.cjs +0 -176
  230. package/dist/lib/auto-linker.d.cts +0 -26
  231. package/dist/lib/config.cjs +0 -78
  232. package/dist/lib/config.d.cts +0 -11
  233. package/dist/lib/entity-index.cjs +0 -84
  234. package/dist/lib/entity-index.d.cts +0 -26
  235. package/dist/lib/project-utils.cjs +0 -864
  236. package/dist/lib/project-utils.d.cts +0 -97
  237. package/dist/lib/session-repair.cjs +0 -239
  238. package/dist/lib/session-repair.d.cts +0 -110
  239. package/dist/lib/session-utils.cjs +0 -209
  240. package/dist/lib/session-utils.d.cts +0 -63
  241. package/dist/lib/task-utils.cjs +0 -1137
  242. package/dist/lib/task-utils.d.cts +0 -208
  243. package/dist/lib/template-engine.cjs +0 -47
  244. package/dist/lib/template-engine.d.cts +0 -11
  245. package/dist/plugin/index.cjs +0 -1907
  246. package/dist/plugin/index.d.cts +0 -36
  247. package/dist/plugin/inject.cjs +0 -356
  248. package/dist/plugin/inject.d.cts +0 -54
  249. package/dist/plugin/inject.d.ts +0 -54
  250. package/dist/plugin/inject.js +0 -17
  251. package/dist/plugin/observe.cjs +0 -631
  252. package/dist/plugin/observe.d.cts +0 -39
  253. package/dist/plugin/observe.d.ts +0 -39
  254. package/dist/plugin/observe.js +0 -18
  255. package/dist/plugin/templates.cjs +0 -593
  256. package/dist/plugin/templates.d.cts +0 -52
  257. package/dist/plugin/templates.d.ts +0 -52
  258. package/dist/plugin/templates.js +0 -25
  259. package/dist/plugin/types.cjs +0 -18
  260. package/dist/plugin/types.d.cts +0 -209
  261. package/dist/plugin/types.d.ts +0 -209
  262. package/dist/plugin/types.js +0 -0
  263. package/dist/plugin/vault.cjs +0 -927
  264. package/dist/plugin/vault.d.cts +0 -68
  265. package/dist/plugin/vault.d.ts +0 -68
  266. package/dist/plugin/vault.js +0 -22
  267. package/dist/types-Y2_Um2Ls.d.cts +0 -205
  268. package/templates/memory-event.md +0 -67
  269. package/templates/party.md +0 -63
  270. package/templates/primitive-registry.yaml +0 -551
  271. package/templates/run.md +0 -68
  272. package/templates/trigger.md +0 -68
  273. package/templates/workspace.md +0 -50
@@ -1,1074 +0,0 @@
1
- import {
2
- DATE_HEADING_RE,
3
- parseObservationMarkdown
4
- } from "./chunk-QK3UCXWL.js";
5
- import {
6
- listObservationFiles
7
- } from "./chunk-Z2XBWN7A.js";
8
-
9
- // src/lib/reweave.ts
10
- import * as fs from "fs";
11
- var SUPERSEDED_MARKER_RE = /\[superseded\|by=([^\]|]+)\|detected=([^\]]+)\]/;
12
- function isSuperseded(line) {
13
- return SUPERSEDED_MARKER_RE.test(line);
14
- }
15
- function getSupersessionInfo(line) {
16
- const m = line.match(SUPERSEDED_MARKER_RE);
17
- if (!m) return null;
18
- return { supersededBy: m[1], detectedAt: m[2] };
19
- }
20
- function makeSupersededMarker(supersedingDate, detectedAt) {
21
- return ` [superseded|by=${supersedingDate}|detected=${detectedAt}]`;
22
- }
23
- function extractEntities(content) {
24
- const normalized = content.toLowerCase().replace(/['']/g, "'");
25
- const quoted = [];
26
- for (const m of normalized.matchAll(/[""]([^""]+)[""]/g)) {
27
- quoted.push(m[1].trim());
28
- }
29
- for (const m of normalized.matchAll(/"([^"]+)"/g)) {
30
- quoted.push(m[1].trim());
31
- }
32
- const patterns = [
33
- /(\w[\w\s]{1,30}?)\s+(?:is|are|was|were|changed to|switched to|moved to|updated to|now uses?|now lives?|now works?)\s+/gi,
34
- /(?:uses?|prefers?|likes?|lives? (?:in|at)|works? (?:at|for)|drives?|owns?)\s+([\w\s]{2,30})/gi,
35
- /(\w[\w\s]{1,20}?)'s\s+(\w[\w\s]{1,20})/gi
36
- ];
37
- const phrases = [...quoted];
38
- for (const pat of patterns) {
39
- for (const m of content.matchAll(pat)) {
40
- if (m[1]) phrases.push(m[1].trim().toLowerCase());
41
- if (m[2]) phrases.push(m[2].trim().toLowerCase());
42
- }
43
- }
44
- const stopwords = /* @__PURE__ */ new Set([
45
- "the",
46
- "a",
47
- "an",
48
- "is",
49
- "are",
50
- "was",
51
- "were",
52
- "be",
53
- "been",
54
- "being",
55
- "have",
56
- "has",
57
- "had",
58
- "do",
59
- "does",
60
- "did",
61
- "will",
62
- "would",
63
- "could",
64
- "should",
65
- "may",
66
- "might",
67
- "shall",
68
- "can",
69
- "need",
70
- "dare",
71
- "ought",
72
- "used",
73
- "to",
74
- "of",
75
- "in",
76
- "for",
77
- "on",
78
- "with",
79
- "at",
80
- "by",
81
- "from",
82
- "as",
83
- "into",
84
- "through",
85
- "during",
86
- "before",
87
- "after",
88
- "above",
89
- "below",
90
- "between",
91
- "out",
92
- "off",
93
- "over",
94
- "under",
95
- "again",
96
- "further",
97
- "then",
98
- "once",
99
- "and",
100
- "but",
101
- "or",
102
- "nor",
103
- "not",
104
- "so",
105
- "yet",
106
- "both",
107
- "either",
108
- "neither",
109
- "each",
110
- "every",
111
- "all",
112
- "any",
113
- "few",
114
- "more",
115
- "most",
116
- "other",
117
- "some",
118
- "such",
119
- "no",
120
- "only",
121
- "own",
122
- "same",
123
- "than",
124
- "too",
125
- "very",
126
- "just",
127
- "because",
128
- "that",
129
- "this",
130
- "these",
131
- "those",
132
- "it",
133
- "its",
134
- "he",
135
- "she",
136
- "they",
137
- "we",
138
- "you",
139
- "i",
140
- "me",
141
- "my",
142
- "his",
143
- "her",
144
- "our",
145
- "your",
146
- "their",
147
- "pedro",
148
- "clawdious"
149
- ]);
150
- const words = normalized.replace(/[^a-z0-9\s'-]/g, " ").split(/\s+/).filter((w) => w.length > 2 && !stopwords.has(w));
151
- return [.../* @__PURE__ */ new Set([...phrases, ...words])].filter(Boolean);
152
- }
153
- function entitySimilarity(a, b) {
154
- if (a.length === 0 || b.length === 0) return 0;
155
- const setA = new Set(a);
156
- const setB = new Set(b);
157
- let overlap = 0;
158
- for (const item of setA) {
159
- if (setB.has(item)) overlap++;
160
- }
161
- const union = (/* @__PURE__ */ new Set([...a, ...b])).size;
162
- return union > 0 ? overlap / union : 0;
163
- }
164
- function isKnowledgeUpdate(older, newer, threshold = 0.3) {
165
- const updateableTypes = /* @__PURE__ */ new Set(["fact", "preference", "decision", "commitment", "project", "relationship"]);
166
- if (!updateableTypes.has(older.type) && !updateableTypes.has(newer.type)) {
167
- return { isUpdate: false, reason: "non-updateable types" };
168
- }
169
- const olderEntities = extractEntities(older.content);
170
- const newerEntities = extractEntities(newer.content);
171
- const similarity = entitySimilarity(olderEntities, newerEntities);
172
- if (similarity < threshold) {
173
- return { isUpdate: false, reason: `low entity similarity: ${similarity.toFixed(2)}` };
174
- }
175
- const normalizeContent = (s) => s.toLowerCase().replace(/\s+/g, " ").trim();
176
- if (normalizeContent(older.content) === normalizeContent(newer.content)) {
177
- return { isUpdate: false, reason: "identical content" };
178
- }
179
- return {
180
- isUpdate: true,
181
- reason: `entity overlap ${similarity.toFixed(2)}: entities=[${olderEntities.slice(0, 3).join(", ")}]`
182
- };
183
- }
184
- function loadObservations(vaultPath, since) {
185
- const files = listObservationFiles(vaultPath, {
186
- fromDate: since
187
- });
188
- return files.map((f) => {
189
- const content = fs.readFileSync(f.path, "utf-8");
190
- const records = parseObservationMarkdown(content);
191
- return { file: f, records };
192
- });
193
- }
194
- function reweave(options) {
195
- const { vaultPath, since, dryRun = false, similarityThreshold = 0.3 } = options;
196
- const allObsFiles = loadObservations(vaultPath);
197
- const newObsFiles = since ? allObsFiles.filter((f) => f.file.date >= since) : allObsFiles;
198
- const allRecordsWithFile = [];
199
- for (const { file, records } of allObsFiles) {
200
- for (let i = 0; i < records.length; i++) {
201
- allRecordsWithFile.push({ record: records[i], file, lineIndex: i });
202
- }
203
- }
204
- allRecordsWithFile.sort((a, b) => a.record.date.localeCompare(b.record.date));
205
- const supersessions = [];
206
- const detectedAt = (/* @__PURE__ */ new Date()).toISOString().slice(0, 10);
207
- for (const { file: newFile, records: newRecords } of newObsFiles) {
208
- for (const newRec of newRecords) {
209
- if (isSuperseded(newRec.rawLine)) continue;
210
- for (const { record: oldRec, file: oldFile } of allRecordsWithFile) {
211
- if (oldRec.date >= newRec.date && oldFile.path === newFile.path) continue;
212
- if (oldRec.date > newRec.date) continue;
213
- if (isSuperseded(oldRec.rawLine)) continue;
214
- const { isUpdate, reason } = isKnowledgeUpdate(oldRec, newRec, similarityThreshold);
215
- if (isUpdate) {
216
- supersessions.push({
217
- oldObservation: oldRec,
218
- newObservation: newRec,
219
- oldFile: oldFile.path,
220
- newFile: newFile.path,
221
- reason,
222
- detectedAt
223
- });
224
- }
225
- }
226
- }
227
- }
228
- if (!dryRun && supersessions.length > 0) {
229
- applySupersessions(supersessions, detectedAt);
230
- }
231
- return {
232
- filesScanned: allObsFiles.length,
233
- observationsChecked: allRecordsWithFile.length,
234
- supersessions,
235
- dryRun
236
- };
237
- }
238
- function applySupersessions(supersessions, detectedAt) {
239
- const byFile = /* @__PURE__ */ new Map();
240
- for (const s of supersessions) {
241
- const existing = byFile.get(s.oldFile) ?? [];
242
- existing.push(s);
243
- byFile.set(s.oldFile, existing);
244
- }
245
- for (const [filePath, records] of byFile) {
246
- let content = fs.readFileSync(filePath, "utf-8");
247
- for (const s of records) {
248
- const oldLine = s.oldObservation.rawLine;
249
- if (content.includes(oldLine) && !isSuperseded(oldLine)) {
250
- const marker = makeSupersededMarker(s.newObservation.date, detectedAt);
251
- content = content.replace(oldLine, oldLine + marker);
252
- }
253
- }
254
- fs.writeFileSync(filePath, content, "utf-8");
255
- }
256
- }
257
- function filterSuperseded(lines) {
258
- return lines.filter((line) => !isSuperseded(line));
259
- }
260
- function stripSupersededObservations(markdown) {
261
- const lines = markdown.split("\n");
262
- const result = [];
263
- for (const line of lines) {
264
- if (DATE_HEADING_RE.test(line) || line.trim() === "") {
265
- result.push(line);
266
- continue;
267
- }
268
- if (isSuperseded(line)) continue;
269
- result.push(line);
270
- }
271
- return result.join("\n").replace(/\n{3,}/g, "\n\n").trim();
272
- }
273
-
274
- // src/lib/search.ts
275
- import { execFileSync, spawnSync } from "child_process";
276
- import * as fs2 from "fs";
277
- import * as path from "path";
278
- var QMD_INSTALL_URL = "https://github.com/tobi/qmd";
279
- var QMD_INSTALL_COMMAND = "bun install -g github:tobi/qmd";
280
- var QMD_NOT_INSTALLED_MESSAGE = `ClawVault requires qmd. Install: ${QMD_INSTALL_COMMAND}`;
281
- var QMD_INDEX_ENV_VAR = "CLAWVAULT_QMD_INDEX";
282
- var QmdUnavailableError = class extends Error {
283
- constructor(message = QMD_NOT_INSTALLED_MESSAGE) {
284
- super(message);
285
- this.name = "QmdUnavailableError";
286
- }
287
- };
288
- function ensureJsonArgs(args) {
289
- return args.includes("--json") ? args : [...args, "--json"];
290
- }
291
- function resolveQmdIndexName(indexName) {
292
- const explicit = indexName?.trim();
293
- if (explicit) {
294
- return explicit;
295
- }
296
- const fromEnv = process.env[QMD_INDEX_ENV_VAR]?.trim();
297
- return fromEnv || void 0;
298
- }
299
- function withQmdIndexArgs(args, indexName) {
300
- if (args.includes("--index")) {
301
- return [...args];
302
- }
303
- const resolvedIndexName = resolveQmdIndexName(indexName);
304
- if (!resolvedIndexName) {
305
- return [...args];
306
- }
307
- return ["--index", resolvedIndexName, ...args];
308
- }
309
- function tryParseJson(raw) {
310
- try {
311
- return JSON.parse(raw);
312
- } catch {
313
- return null;
314
- }
315
- }
316
- function extractJsonPayload(raw) {
317
- const start = raw.search(/[\[{]/);
318
- if (start === -1) return null;
319
- const end = Math.max(raw.lastIndexOf("]"), raw.lastIndexOf("}"));
320
- if (end <= start) return null;
321
- return raw.slice(start, end + 1);
322
- }
323
- function stripQmdNoise(raw) {
324
- return raw.split("\n").filter((line) => {
325
- const t = line.trim();
326
- if (!t) return true;
327
- if (t.startsWith("[node-llama-cpp]")) return false;
328
- if (t.startsWith("Expanding query")) return false;
329
- if (t.startsWith("Searching ") && t.endsWith("queries...")) return false;
330
- if (/^[├└─│]/.test(t)) return false;
331
- return true;
332
- }).join("\n");
333
- }
334
- function parseQmdOutput(raw) {
335
- const trimmed = stripQmdNoise(raw).trim();
336
- if (!trimmed) return [];
337
- if (trimmed.startsWith("No results") || trimmed.startsWith("No matches")) return [];
338
- const direct = tryParseJson(trimmed);
339
- const extracted = direct ? null : extractJsonPayload(trimmed);
340
- const parsed = direct ?? (extracted ? tryParseJson(extracted) : null);
341
- if (!parsed) {
342
- throw new Error("qmd returned non-JSON output. Ensure qmd supports --json.");
343
- }
344
- if (Array.isArray(parsed)) {
345
- return parsed;
346
- }
347
- if (parsed && typeof parsed === "object") {
348
- const candidate = parsed.results ?? parsed.items ?? parsed.data;
349
- if (Array.isArray(candidate)) {
350
- return candidate;
351
- }
352
- }
353
- throw new Error("qmd returned an unexpected JSON shape.");
354
- }
355
- function ensureQmdAvailable() {
356
- if (!hasQmd()) {
357
- throw new QmdUnavailableError();
358
- }
359
- }
360
- function execQmd(args, indexName) {
361
- ensureQmdAvailable();
362
- const finalArgs = withQmdIndexArgs(ensureJsonArgs(args), indexName);
363
- try {
364
- const result = execFileSync("qmd", finalArgs, {
365
- encoding: "utf-8",
366
- stdio: ["ignore", "pipe", "pipe"],
367
- maxBuffer: 10 * 1024 * 1024
368
- // 10MB
369
- });
370
- return parseQmdOutput(result);
371
- } catch (err) {
372
- if (err?.code === "ENOENT") {
373
- throw new QmdUnavailableError();
374
- }
375
- const output = [err?.stdout, err?.stderr].filter(Boolean).join("\n");
376
- if (output) {
377
- try {
378
- return parseQmdOutput(output);
379
- } catch {
380
- }
381
- }
382
- const message = err?.message ? `qmd failed: ${err.message}` : "qmd failed";
383
- throw new Error(message);
384
- }
385
- }
386
- function hasQmd() {
387
- const result = spawnSync("qmd", ["--version"], { stdio: "ignore" });
388
- return !result.error;
389
- }
390
- function qmdUpdate(collection, indexName) {
391
- ensureQmdAvailable();
392
- const args = ["update"];
393
- if (collection) {
394
- args.push("-c", collection);
395
- }
396
- execFileSync("qmd", withQmdIndexArgs(args, indexName), { stdio: "inherit" });
397
- }
398
- function qmdEmbed(collection, indexName) {
399
- ensureQmdAvailable();
400
- const args = ["embed"];
401
- if (collection) {
402
- args.push("-c", collection);
403
- }
404
- execFileSync("qmd", withQmdIndexArgs(args, indexName), { stdio: "inherit" });
405
- }
406
- function sentenceChunk(text, maxChars = 600, overlapSentences = 1) {
407
- const sentences = text.split(/(?<=[.!?])\s+|\n{2,}/).map((s) => s.trim()).filter(Boolean);
408
- if (sentences.length === 0) return text.trim() ? [text] : [];
409
- const chunks = [];
410
- let i = 0;
411
- while (i < sentences.length) {
412
- const chunkSents = [];
413
- let chunkLen = 0;
414
- let j = i;
415
- while (j < sentences.length && chunkLen + sentences[j].length < maxChars) {
416
- chunkSents.push(sentences[j]);
417
- chunkLen += sentences[j].length + 1;
418
- j++;
419
- }
420
- if (chunkSents.length === 0) {
421
- chunkSents.push(sentences[j].slice(0, maxChars));
422
- j++;
423
- }
424
- chunks.push(chunkSents.join(" "));
425
- i = Math.max(j - overlapSentences, i + 1);
426
- }
427
- return chunks;
428
- }
429
- var STOPWORDS = /* @__PURE__ */ new Set([
430
- "what",
431
- "when",
432
- "where",
433
- "which",
434
- "that",
435
- "this",
436
- "have",
437
- "from",
438
- "with",
439
- "they",
440
- "been",
441
- "were",
442
- "will",
443
- "about",
444
- "would",
445
- "could",
446
- "should",
447
- "their",
448
- "there",
449
- "does",
450
- "your",
451
- "more",
452
- "some",
453
- "than",
454
- "into",
455
- "also",
456
- "just",
457
- "very",
458
- "much",
459
- "most",
460
- "many",
461
- "only",
462
- "other",
463
- "each",
464
- "every",
465
- "after",
466
- "before",
467
- "did",
468
- "the",
469
- "and",
470
- "for",
471
- "are",
472
- "was",
473
- "not",
474
- "but",
475
- "can",
476
- "had",
477
- "has",
478
- "how",
479
- "who",
480
- "why",
481
- "its",
482
- "you",
483
- "my",
484
- "me",
485
- "is",
486
- "it",
487
- "do",
488
- "so",
489
- "if",
490
- "or",
491
- "an",
492
- "on",
493
- "at",
494
- "by",
495
- "no",
496
- "up",
497
- "to",
498
- "in",
499
- "of",
500
- "am",
501
- "be"
502
- ]);
503
- function tokenize(text) {
504
- return text.toLowerCase().split(/\s+/).map((w) => w.replace(/^[?.,!"'\-():;[\]{}*]+|[?.,!"'\-():;[\]{}*]+$/g, "")).filter((w) => w.length > 1);
505
- }
506
- function queryTerms(query) {
507
- return tokenize(query).filter((w) => !STOPWORDS.has(w));
508
- }
509
- function bm25RankChunks(chunks, terms, max = 5) {
510
- if (chunks.length === 0) return [];
511
- const termSet = new Set(terms);
512
- const scored = chunks.map((text, idx) => {
513
- const words = new Set(tokenize(text));
514
- let overlap = 0;
515
- for (const t of termSet) if (words.has(t)) overlap++;
516
- return { text, score: overlap, idx };
517
- });
518
- scored.sort((a, b) => b.score - a.score);
519
- const seen = /* @__PURE__ */ new Set();
520
- const result = [];
521
- seen.add(0);
522
- result.push({ text: chunks[0], score: scored.find((s) => s.idx === 0)?.score ?? 0 });
523
- for (const s of scored) {
524
- if (result.length >= max) break;
525
- if (!seen.has(s.idx) && s.score > 0) {
526
- seen.add(s.idx);
527
- result.push({ text: s.text, score: s.score });
528
- }
529
- }
530
- return result;
531
- }
532
- var MONTH_NAMES = {
533
- january: 1,
534
- february: 2,
535
- march: 3,
536
- april: 4,
537
- may: 5,
538
- june: 6,
539
- july: 7,
540
- august: 8,
541
- september: 9,
542
- october: 10,
543
- november: 11,
544
- december: 12,
545
- jan: 1,
546
- feb: 2,
547
- mar: 3,
548
- apr: 4,
549
- jun: 6,
550
- jul: 7,
551
- aug: 8,
552
- sep: 9,
553
- sept: 9,
554
- oct: 10,
555
- nov: 11,
556
- dec: 12
557
- };
558
- var MONTH_RE_PART = Object.keys(MONTH_NAMES).join("|");
559
- var DATE_ISO_RE = /\b(\d{4})[/-](\d{1,2})[/-](\d{1,2})\b/g;
560
- var DATE_US_RE = /\b(\d{1,2})\/(\d{1,2})\/(\d{4})\b/g;
561
- var DATE_MONTH_DAY_YEAR_RE = new RegExp(
562
- `\\b(${MONTH_RE_PART})\\s+(\\d{1,2})(?:st|nd|rd|th)?,?\\s*(\\d{4})\\b`,
563
- "gi"
564
- );
565
- var DATE_DAY_MONTH_YEAR_RE = new RegExp(
566
- `\\b(\\d{1,2})(?:st|nd|rd|th)?\\s+(${MONTH_RE_PART}),?\\s*(\\d{4})\\b`,
567
- "gi"
568
- );
569
- var DATE_MONTH_DAY_RE = new RegExp(
570
- `\\b(${MONTH_RE_PART})\\s+(\\d{1,2})(?:st|nd|rd|th)?\\b`,
571
- "gi"
572
- );
573
- var RELATIVE_AGO_RE = /\b(\d+)\s+(days?|weeks?|months?|years?)\s+ago\b/gi;
574
- var RELATIVE_IN_RE = /\bin\s+(\d+)\s+(days?|weeks?|months?|years?)\b/gi;
575
- var DURATION_RE = /(?:for|took|spent|lasted|about|approximately|around)\s+(\d+)\s+(days?|weeks?|months?|years?|hours?|minutes?)/gi;
576
- function tryParseISODate(y, m, d) {
577
- const dt = new Date(Date.UTC(y, m - 1, d));
578
- if (dt.getUTCFullYear() === y && dt.getUTCMonth() === m - 1 && dt.getUTCDate() === d) return dt;
579
- return null;
580
- }
581
- function unitToDays(n, unit) {
582
- const u = unit.toLowerCase().replace(/s$/, "");
583
- switch (u) {
584
- case "day":
585
- return n;
586
- case "week":
587
- return n * 7;
588
- case "month":
589
- return n * 30;
590
- case "year":
591
- return n * 365;
592
- default:
593
- return null;
594
- }
595
- }
596
- function contextSnippet(text, start, end, maxLen = 150) {
597
- const s = Math.max(0, start - Math.floor(maxLen / 2));
598
- const e = Math.min(text.length, end + Math.floor(maxLen / 2));
599
- return text.slice(s, e).replace(/\n/g, " ").trim();
600
- }
601
- function isoStr(d) {
602
- const yy = d.getUTCFullYear();
603
- const mm = String(d.getUTCMonth() + 1).padStart(2, "0");
604
- const dd = String(d.getUTCDate()).padStart(2, "0");
605
- return `${yy}-${mm}-${dd}`;
606
- }
607
- function extractDates(text, sessionDateStr) {
608
- const results = [];
609
- const sessionDate = sessionDateStr ? new Date(sessionDateStr) : null;
610
- const seen = /* @__PURE__ */ new Set();
611
- function push(date, ctx, docId = "") {
612
- const key = `${date}|${ctx.slice(0, 60)}`;
613
- if (seen.has(key)) return;
614
- seen.add(key);
615
- results.push({ date, context: ctx, documentId: docId });
616
- }
617
- for (const m of text.matchAll(DATE_ISO_RE)) {
618
- const dt = tryParseISODate(+m[1], +m[2], +m[3]);
619
- if (dt) push(isoStr(dt), contextSnippet(text, m.index, m.index + m[0].length));
620
- }
621
- for (const m of text.matchAll(DATE_US_RE)) {
622
- const dt = tryParseISODate(+m[3], +m[1], +m[2]);
623
- if (dt) push(isoStr(dt), contextSnippet(text, m.index, m.index + m[0].length));
624
- }
625
- for (const m of text.matchAll(DATE_MONTH_DAY_YEAR_RE)) {
626
- const mon = MONTH_NAMES[m[1].toLowerCase()];
627
- if (mon) {
628
- const dt = tryParseISODate(+m[3], mon, +m[2]);
629
- if (dt) push(isoStr(dt), contextSnippet(text, m.index, m.index + m[0].length));
630
- }
631
- }
632
- for (const m of text.matchAll(DATE_DAY_MONTH_YEAR_RE)) {
633
- const mon = MONTH_NAMES[m[2].toLowerCase()];
634
- if (mon) {
635
- const dt = tryParseISODate(+m[3], mon, +m[1]);
636
- if (dt) push(isoStr(dt), contextSnippet(text, m.index, m.index + m[0].length));
637
- }
638
- }
639
- if (sessionDate) {
640
- for (const m of text.matchAll(DATE_MONTH_DAY_RE)) {
641
- const mon = MONTH_NAMES[m[1].toLowerCase()];
642
- if (mon) {
643
- const dt = tryParseISODate(sessionDate.getFullYear(), mon, +m[2]);
644
- if (dt) push(isoStr(dt), contextSnippet(text, m.index, m.index + m[0].length));
645
- }
646
- }
647
- }
648
- if (sessionDate) {
649
- for (const m of text.matchAll(RELATIVE_AGO_RE)) {
650
- const days = unitToDays(+m[1], m[2]);
651
- if (days !== null) {
652
- const dt = new Date(sessionDate.getTime() - days * 864e5);
653
- push(isoStr(dt), contextSnippet(text, m.index, m.index + m[0].length));
654
- }
655
- }
656
- for (const m of text.matchAll(RELATIVE_IN_RE)) {
657
- const days = unitToDays(+m[1], m[2]);
658
- if (days !== null) {
659
- const dt = new Date(sessionDate.getTime() + days * 864e5);
660
- push(isoStr(dt), contextSnippet(text, m.index, m.index + m[0].length));
661
- }
662
- }
663
- }
664
- for (const m of text.matchAll(DURATION_RE)) {
665
- push(`duration:${m[1]} ${m[2]}`, contextSnippet(text, m.index, m.index + m[0].length));
666
- }
667
- return results;
668
- }
669
- var PREF_PATTERNS = [
670
- // "I use/prefer/like/love/enjoy X"
671
- /\bi\s+(?:use|prefer|like|love|enjoy|favor|chose|switched to|started using|always use|usually use)\s+(.{3,60}?)(?:[.,;!?\n]|$)/gi,
672
- // "my favorite X is Y"
673
- /\bmy\s+(?:favorite|preferred|go-to|usual)\s+\w+\s+(?:is|are|was)\s+(.{3,60}?)(?:[.,;!?\n]|$)/gi,
674
- // "I'm a big fan of X"
675
- /\bi(?:'m| am)\s+(?:a )?(?:big |huge )?fan of\s+(.{3,60}?)(?:[.,;!?\n]|$)/gi,
676
- // "I switched from X to Y"
677
- /\bi\s+switched\s+from\s+(.{3,40}?)\s+to\s+(.{3,40}?)(?:[.,;!?\n]|$)/gi
678
- ];
679
- function extractPreferences(text, documentId = "") {
680
- const results = [];
681
- const seen = /* @__PURE__ */ new Set();
682
- for (const pattern of PREF_PATTERNS) {
683
- for (const m of text.matchAll(pattern)) {
684
- const value = (m[1] || "").trim();
685
- if (!value || value.length < 3) continue;
686
- const key = value.toLowerCase();
687
- if (seen.has(key)) continue;
688
- seen.add(key);
689
- const ctx = contextSnippet(text, m.index, m.index + m[0].length, 200);
690
- let category = "general";
691
- if (/tool|software|app|editor|ide|framework|library|language/i.test(ctx)) category = "tool";
692
- else if (/hobby|sport|exercise|game|play/i.test(ctx)) category = "hobby";
693
- else if (/brand|product|model|device|hardware/i.test(ctx)) category = "brand";
694
- else if (/food|drink|restaurant|cuisine|recipe/i.test(ctx)) category = "food";
695
- else if (/music|movie|show|book|podcast|artist|band/i.test(ctx)) category = "entertainment";
696
- results.push({ category, value, documentId, context: ctx });
697
- }
698
- }
699
- return results;
700
- }
701
- var PREFERENCE_Q_RE = /(?:can you (?:recommend|suggest)|any (?:tips|advice|suggestions|recommendations)|what .*(?:recommend|suggest)|what should i|where should i|which .* should i|please (?:recommend|suggest)|based on .* (?:interest|preference|taste)|personalized|tailored to (?:my|me))/i;
702
- var TEMPORAL_Q_RE = /(?:how many (?:days|weeks|months|years|hours|minutes) (?:passed|did|have|ago|between|since|in total|took)|how long (?:did|was|were|have|has|does)|how long ago|what (?:is the )?order|in order|which .* (?:first|last|earlier|later|before|after|most recent|oldest|newest)|chronological|(?:earlier|later|sooner|newer|older) than)/i;
703
- var AGGREGATION_Q_RE = /(?:how many|how much|total|all the|count|list all|every|what are all|name all)/i;
704
- function classifyQuestion(q) {
705
- if (PREFERENCE_Q_RE.test(q)) return "preference";
706
- if (TEMPORAL_Q_RE.test(q)) return "temporal";
707
- if (!TEMPORAL_Q_RE.test(q) && AGGREGATION_Q_RE.test(q)) return "aggregation";
708
- return "default";
709
- }
710
- var SearchEngine = class {
711
- documents = /* @__PURE__ */ new Map();
712
- collection = "clawvault";
713
- vaultPath = "";
714
- collectionRoot = "";
715
- qmdIndexName;
716
- /** v2.7 — Per-document date index built at ingest time */
717
- dateIndex = /* @__PURE__ */ new Map();
718
- /** v2.7 — Per-document preference index built at ingest time */
719
- preferenceIndex = /* @__PURE__ */ new Map();
720
- /** v2.7 — Per-document chunk cache for BM25 pre-filtering */
721
- chunkCache = /* @__PURE__ */ new Map();
722
- /**
723
- * Set the collection name (usually vault name)
724
- */
725
- setCollection(name) {
726
- this.collection = name;
727
- }
728
- /**
729
- * Set the vault path for file resolution
730
- */
731
- setVaultPath(vaultPath) {
732
- this.vaultPath = vaultPath;
733
- }
734
- /**
735
- * Set the collection root for qmd:// URI resolution
736
- */
737
- setCollectionRoot(root) {
738
- this.collectionRoot = path.resolve(root);
739
- }
740
- /**
741
- * Set qmd index name (defaults to qmd global default when omitted)
742
- */
743
- setIndexName(indexName) {
744
- this.qmdIndexName = indexName;
745
- }
746
- /**
747
- * Add or update a document in the local cache.
748
- * v2.7: also extracts dates, preferences, and chunks at ingest time.
749
- * Note: qmd indexing happens via qmd update command
750
- */
751
- addDocument(doc) {
752
- this.documents.set(doc.id, doc);
753
- if (doc.content) {
754
- const sessionDate = doc.modified ? isoStr(doc.modified) : void 0;
755
- const dates = extractDates(doc.content, sessionDate);
756
- for (const d of dates) d.documentId = doc.id;
757
- if (dates.length > 0) this.dateIndex.set(doc.id, dates);
758
- const prefs = extractPreferences(doc.content, doc.id);
759
- if (prefs.length > 0) this.preferenceIndex.set(doc.id, prefs);
760
- const chunks = sentenceChunk(doc.content, 600, 1);
761
- if (chunks.length > 0) this.chunkCache.set(doc.id, chunks);
762
- }
763
- }
764
- /**
765
- * Remove a document from the local cache
766
- */
767
- removeDocument(id) {
768
- this.documents.delete(id);
769
- this.dateIndex.delete(id);
770
- this.preferenceIndex.delete(id);
771
- this.chunkCache.delete(id);
772
- }
773
- /**
774
- * No-op for qmd - indexing is managed externally
775
- */
776
- rebuildIDF() {
777
- }
778
- /**
779
- * BM25 search via qmd
780
- */
781
- search(query, options = {}) {
782
- return this.runQmdQuery("search", query, options);
783
- }
784
- /**
785
- * Vector/semantic search via qmd vsearch
786
- */
787
- vsearch(query, options = {}) {
788
- return this.runQmdQuery("vsearch", query, options);
789
- }
790
- /**
791
- * Combined search with query expansion (qmd query command)
792
- */
793
- query(query, options = {}) {
794
- return this.runQmdQuery("query", query, options);
795
- }
796
- runQmdQuery(command, query, options) {
797
- const {
798
- limit = 10,
799
- minScore = 0,
800
- category,
801
- tags,
802
- fullContent = false,
803
- temporalBoost = false,
804
- relevanceThreshold,
805
- thresholdMaxResults = 40
806
- } = options;
807
- if (!query.trim()) return [];
808
- const fetchLimit = relevanceThreshold !== void 0 ? thresholdMaxResults * 2 : limit * 2;
809
- const args = [
810
- command,
811
- query,
812
- "-n",
813
- String(fetchLimit),
814
- "--json"
815
- ];
816
- if (this.collection) {
817
- args.push("-c", this.collection);
818
- }
819
- const qmdResults = execQmd(args, this.qmdIndexName);
820
- const effectiveLimit = relevanceThreshold !== void 0 ? thresholdMaxResults : limit;
821
- const results = this.convertResults(qmdResults, {
822
- limit: effectiveLimit,
823
- minScore: relevanceThreshold !== void 0 ? relevanceThreshold : minScore,
824
- category,
825
- tags,
826
- fullContent,
827
- temporalBoost
828
- });
829
- return results;
830
- }
831
- // -------------------------------------------------------------------------
832
- // v2.7 — New public APIs
833
- // -------------------------------------------------------------------------
834
- /**
835
- * v2.7 — Chunk-level BM25 pre-filtered search. Ranks chunks within each
836
- * document by keyword relevance before semantic ranking, so relevant
837
- * content deep in long documents isn't missed.
838
- *
839
- * Returns results with snippets from the best-matching chunks.
840
- */
841
- chunkPrefilterSearch(query, options = {}) {
842
- const terms = queryTerms(query);
843
- const results = this.runQmdQuery("query", query, options);
844
- for (const r of results) {
845
- const chunks = this.chunkCache.get(r.document.id);
846
- if (chunks && chunks.length > 0 && terms.length > 0) {
847
- const ranked = bm25RankChunks(chunks, terms, 3);
848
- if (ranked.length > 0 && ranked[0].score > 0) {
849
- r.snippet = ranked.map((c) => c.text).join("\n...\n").slice(0, 600);
850
- }
851
- }
852
- }
853
- return results;
854
- }
855
- /**
856
- * v2.7 — Exhaustive threshold-based search for aggregation queries.
857
- * Keeps pulling results until relevance drops below threshold.
858
- */
859
- exhaustiveSearch(query, threshold = 0.01, maxResults = 40) {
860
- return this.runQmdQuery("query", query, {
861
- relevanceThreshold: threshold,
862
- thresholdMaxResults: maxResults,
863
- fullContent: false
864
- });
865
- }
866
- /**
867
- * v2.7 — Get all extracted dates, optionally filtered by document ids.
868
- */
869
- getDates(documentIds) {
870
- const all = [];
871
- const iter = documentIds ? documentIds.map((id) => [id, this.dateIndex.get(id)]).filter(([, v]) => v) : this.dateIndex.entries();
872
- for (const [, dates] of iter) {
873
- if (dates) all.push(...dates);
874
- }
875
- return all;
876
- }
877
- /**
878
- * v2.7 — Get all extracted preferences, optionally filtered by document ids.
879
- */
880
- getPreferences(documentIds) {
881
- const all = [];
882
- const iter = documentIds ? documentIds.map((id) => [id, this.preferenceIndex.get(id)]).filter(([, v]) => v) : this.preferenceIndex.entries();
883
- for (const [, prefs] of iter) {
884
- if (prefs) all.push(...prefs);
885
- }
886
- return all;
887
- }
888
- /**
889
- * v2.7 — Search with automatic strategy selection based on question type.
890
- * Classifies the query and routes to the appropriate pipeline.
891
- */
892
- smartQuery(query, options = {}) {
893
- const qtype = classifyQuestion(query);
894
- switch (qtype) {
895
- case "aggregation":
896
- return this.exhaustiveSearch(query, 0.01, options.thresholdMaxResults ?? 40);
897
- case "preference":
898
- case "temporal":
899
- default:
900
- return this.chunkPrefilterSearch(query, { ...options, limit: options.limit ?? 10 });
901
- }
902
- }
903
- /**
904
- * Convert qmd results to ClawVault SearchResult format
905
- */
906
- convertResults(qmdResults, options) {
907
- const { limit = 10, minScore = 0, category, tags, fullContent = false, temporalBoost = false } = options;
908
- const results = [];
909
- const maxScore = qmdResults[0]?.score || 1;
910
- for (const qr of qmdResults) {
911
- const filePath = this.qmdUriToPath(qr.file);
912
- const relativePath = this.vaultPath ? path.relative(this.vaultPath, filePath) : filePath;
913
- const normalizedRelativePath = relativePath.replace(/\\/g, "/");
914
- if (normalizedRelativePath.startsWith("ledger/archive/") || normalizedRelativePath.includes("/ledger/archive/")) {
915
- continue;
916
- }
917
- const docId = normalizedRelativePath.replace(/\.md$/, "");
918
- let doc = this.documents.get(docId) ?? this.documents.get(docId.split("/").join(path.sep));
919
- const modifiedAt = this.resolveModifiedAt(doc, filePath);
920
- const parts = normalizedRelativePath.split("/");
921
- const docCategory = parts.length > 1 ? parts[0] : "root";
922
- if (category && docCategory !== category) continue;
923
- if (tags && tags.length > 0 && doc) {
924
- const docTags = new Set(doc.tags);
925
- if (!tags.some((t) => docTags.has(t))) continue;
926
- }
927
- const normalizedScore = maxScore > 0 ? qr.score / maxScore : 0;
928
- const finalScore = temporalBoost ? normalizedScore * this.getRecencyFactor(modifiedAt) : normalizedScore;
929
- if (finalScore < minScore) continue;
930
- if (!doc) {
931
- doc = {
932
- id: docId,
933
- path: filePath,
934
- category: docCategory,
935
- title: qr.title || path.basename(relativePath, ".md"),
936
- content: "",
937
- // Content loaded separately if needed
938
- frontmatter: {},
939
- links: [],
940
- tags: [],
941
- modified: modifiedAt
942
- };
943
- }
944
- results.push({
945
- document: fullContent ? doc : { ...doc, content: "" },
946
- score: finalScore,
947
- snippet: this.stripSupersededFromSnippet(this.cleanSnippet(qr.snippet)),
948
- matchedTerms: []
949
- // qmd doesn't provide this
950
- });
951
- }
952
- return results.sort((a, b) => b.score - a.score).slice(0, limit);
953
- }
954
- resolveModifiedAt(doc, filePath) {
955
- if (doc) return doc.modified;
956
- try {
957
- return fs2.statSync(filePath).mtime;
958
- } catch {
959
- return /* @__PURE__ */ new Date(0);
960
- }
961
- }
962
- getRecencyFactor(modifiedAt) {
963
- const ageMs = Math.max(0, Date.now() - modifiedAt.getTime());
964
- const ageDays = ageMs / (24 * 60 * 60 * 1e3);
965
- if (ageDays < 1) return 1;
966
- if (ageDays <= 7) return 0.9;
967
- return 0.7;
968
- }
969
- /**
970
- * Convert qmd:// URI to file path
971
- */
972
- qmdUriToPath(uri) {
973
- if (uri.startsWith("qmd://")) {
974
- const withoutScheme = uri.slice(6);
975
- const slashIndex = withoutScheme.indexOf("/");
976
- if (slashIndex > -1) {
977
- const relativePath = withoutScheme.slice(slashIndex + 1);
978
- const root = this.collectionRoot || this.vaultPath;
979
- if (root) {
980
- return path.join(root, relativePath);
981
- }
982
- return relativePath;
983
- }
984
- }
985
- return uri;
986
- }
987
- /**
988
- * v2.8 — Filter superseded observation lines from snippet text.
989
- * Ensures search results prefer the latest version of knowledge.
990
- */
991
- stripSupersededFromSnippet(snippet) {
992
- if (!snippet) return snippet;
993
- return snippet.split("\n").filter((line) => !isSuperseded(line)).join("\n");
994
- }
995
- /**
996
- * Clean up qmd snippet format
997
- */
998
- cleanSnippet(snippet) {
999
- if (!snippet) return "";
1000
- return snippet.replace(/@@ [-+]?\d+,?\d* @@ \([^)]+\)/g, "").trim().split("\n").slice(0, 3).join("\n").slice(0, 300);
1001
- }
1002
- /**
1003
- * Get all cached documents
1004
- */
1005
- getAllDocuments() {
1006
- return [...this.documents.values()];
1007
- }
1008
- /**
1009
- * Get document count
1010
- */
1011
- get size() {
1012
- return this.documents.size;
1013
- }
1014
- /**
1015
- * Clear the local document cache and all v2.7 indices
1016
- */
1017
- clear() {
1018
- this.documents.clear();
1019
- this.dateIndex.clear();
1020
- this.preferenceIndex.clear();
1021
- this.chunkCache.clear();
1022
- }
1023
- /**
1024
- * Export documents for persistence
1025
- */
1026
- export() {
1027
- return {
1028
- documents: [...this.documents.values()]
1029
- };
1030
- }
1031
- /**
1032
- * Import from persisted data
1033
- */
1034
- import(data) {
1035
- this.clear();
1036
- for (const doc of data.documents) {
1037
- this.addDocument(doc);
1038
- }
1039
- }
1040
- };
1041
- function extractWikiLinks(content) {
1042
- const matches = content.match(/\[\[([^\]]+)\]\]/g) || [];
1043
- return matches.map((m) => m.slice(2, -2).toLowerCase());
1044
- }
1045
- function extractTags(content) {
1046
- const matches = content.match(/#[\w-]+/g) || [];
1047
- return [...new Set(matches.map((m) => m.slice(1).toLowerCase()))];
1048
- }
1049
-
1050
- export {
1051
- isSuperseded,
1052
- getSupersessionInfo,
1053
- extractEntities,
1054
- entitySimilarity,
1055
- isKnowledgeUpdate,
1056
- reweave,
1057
- filterSuperseded,
1058
- stripSupersededObservations,
1059
- QMD_INSTALL_URL,
1060
- QMD_INSTALL_COMMAND,
1061
- QmdUnavailableError,
1062
- withQmdIndexArgs,
1063
- hasQmd,
1064
- qmdUpdate,
1065
- qmdEmbed,
1066
- sentenceChunk,
1067
- bm25RankChunks,
1068
- extractDates,
1069
- extractPreferences,
1070
- classifyQuestion,
1071
- SearchEngine,
1072
- extractWikiLinks,
1073
- extractTags
1074
- };