sensemaking 0.15.3 → 0.16.0

package/dist/esm/db/open.js.map

@@ -1 +1 @@
-{"version":3,"sources":["/Users/kevin/Dev/OpenSource/ai/sensemaking/src/db/open.ts"],"sourcesContent":["// The Node floor (>=22.20) is explained here and nowhere else: 22.20 is the first release with\n// both FTS5 and row-returning INSERT ... RETURNING. Raise it only for a load-bearing capability.\nimport { mkdirSync, rmSync } from 'node:fs';\nimport { join } from 'node:path';\nimport { DatabaseSync } from 'node:sqlite';\nimport type { Config, ResolvedConfig } from '../config/index.ts';\nimport { contentTokenize, featureSignature, STATE_DIR } from '../config/index.ts';\nimport { SenseError } from '../errors.ts';\nimport { activeFeatures } from '../features/index.ts';\nimport { registerFunctions } from '../sql-functions.ts';\nimport { changedSignatureKeys, rebuildContentTable, reconcile, signatureDiff } from './reconcile.ts';\nimport { getMeta, setMeta } from './shared.ts';\n\nexport const DB_FILENAME = 'cache.db';\n// Cache shape version, independent of the config's own `version`. Bumping it rebuilds\n// existing trees on first query.\nexport const SCHEMA_VERSION = '16';\n\nexport interface OpenResult {\n  db: DatabaseSync;\n  cfg: ResolvedConfig;\n  dbPath: string;\n  parsed: number;\n  warnings: string[];\n}\n\n// Stemming is English-only, but the segmentation underneath it is what decides coverage:\n// unicode61 splits on spaces, so a language written without them (Chinese, Japanese, Thai)\n// indexes a whole run as one token and word search finds nothing. `content.tokenize` is how\n// such a tree picks trigram instead.\nconst DEFAULT_TOKENIZE = 'porter unicode61';\n\n// FTS5 takes its tokenizer as a string literal inside DDL, where nothing can bind, so a\n// configured value has to be concatenated. Probing a throwaway table is what makes that safe\n// and is also the whole validation: anything the linked SQLite accepts passes, anything else\n// fails here with SQLite's own message rather than against the real table. It means no table\n// of which version added which tokenizer has to be maintained.\nfunction resolveTokenize(db: DatabaseSync, cfg: Config): string {\n  const configured = contentTokenize(cfg);\n  if (configured === undefined) return DEFAULT_TOKENIZE;\n  const literal = configured.replace(/'/g, \"''\");\n  try {\n    db.exec('DROP TABLE IF EXISTS temp.sense_tokenize_probe');\n    db.exec(`CREATE VIRTUAL TABLE temp.sense_tokenize_probe USING fts5(x, tokenize = '${literal}')`);\n    db.exec('DROP TABLE IF EXISTS temp.sense_tokenize_probe');\n  } catch (err) {\n    throw new SenseError('CONFIG_INVALID', `content.tokenize \"${configured}\" is not a tokenizer this SQLite accepts (${(err as Error).message}); the built-in choices are unicode61, ascii, porter, and trigram, each with their own options`);\n  }\n  return literal;\n}\n\n// The tokenizer the content table was actually built with, from its own DDL -- the one\n// record that cannot desynchronize from the table. NULL when the table does not exist yet.\nfunction storedTokenize(db: DatabaseSync): string | null {\n  const row = db.prepare(`SELECT sql FROM sqlite_master WHERE name = 'content'`).get() as { sql: string } | undefined;\n  if (!row) return null;\n  const m = row.sql.match(/tokenize = '((?:[^']|'')*)'/);\n  return m ? m[1] : null;\n}\n\n// Content is a separate table (not a column on frontmatter) so `SELECT * FROM frontmatter`\n// can't dump file text into context. Features add their own tables after the core ones.\nfunction ensureSchema(db: DatabaseSync, cfg: Config, tokenize: string): void {\n  db.exec(`CREATE TABLE IF NOT EXISTS frontmatter (\"path\" TEXT PRIMARY KEY, \"_mtime\" REAL, \"_ctime\" REAL, \"_size\" INTEGER, \"_parse_error\" TEXT)`);\n  // IF NOT EXISTS is safe against a tokenizer change: open() compares the table's own DDL\n  // against the resolved tokenizer before this runs, so a stale table is already gone by now.\n  // The three `_seg` sidecars are appended after path, never inserted: bm25(content, ...) and\n  // snippet(content, 2, ...) are documented against the first three columns and keep working\n  // (FTS5 defaults the weights it was not given). Each carries its field's exploded unspaced\n  // runs for text that needs it, and an empty string for text that does not.\n  db.exec(`CREATE VIRTUAL TABLE IF NOT EXISTS content USING fts5(title, summary, text, path UNINDEXED, title_seg, summary_seg, text_seg, tokenize = '${tokenize}')`);\n  // Coverage, not ownership: a path can appear under several presets. path leads the PK so the\n  // per-doc delete is an index hit -- keyed the other way, cold builds went quadratic.\n  db.exec(`CREATE TABLE IF NOT EXISTS preset_files (\"path\" TEXT, preset TEXT, PRIMARY KEY (\"path\", preset))`);\n  db.exec('CREATE INDEX IF NOT EXISTS preset_files_preset ON preset_files(preset)');\n  for (const feature of activeFeatures(cfg)) feature.schema(db);\n  if (getMeta(db, 'schema_version') === null) setMeta(db, 'schema_version', SCHEMA_VERSION);\n  if (getMeta(db, 'features') === null) setMeta(db, 'features', featureSignature(cfg));\n}\n\nexport function docCount(db: DatabaseSync): number {\n  const row = db.prepare('SELECT COUNT(*) AS n FROM frontmatter').get() as { n: number };\n  return row.n;\n}\n\nexport function open(cfg: ResolvedConfig): OpenResult {\n  const stateDir = join(cfg.baseDir, STATE_DIR);\n  mkdirSync(stateDir, { recursive: true });\n  const dbPath = join(stateDir, DB_FILENAME);\n\n  const db = new DatabaseSync(dbPath);\n  db.exec('PRAGMA journal_mode = WAL');\n  // Covers a concurrent watcher's bulk reconcile (~5s for 500 files at 26k notes). A query\n  // that outwaits it still fails loudly.\n  db.exec('PRAGMA busy_timeout = 30000');\n  registerFunctions(db, contentTokenize(cfg) === undefined);\n\n  db.exec('CREATE TABLE IF NOT EXISTS meta (key TEXT PRIMARY KEY, value TEXT)');\n\n  // Before the rebuild branch below, never after: that branch deletes the cache, so a typo'd\n  // tokenizer validated later would cost a full re-index (and re-embed) to reach its own error.\n  const tokenize = resolveTokenize(db, cfg);\n\n  // Schema-version or feature-set mismatch: reconcile only reparses changed files, so an\n  // old cache can't be patched incrementally -- rebuild instead (cheap: nothing expensive lives here).\n  const version = getMeta(db, 'schema_version');\n  const features = getMeta(db, 'features');\n  const wantFeatures = featureSignature(cfg);\n  let tokenizeOnlyRebuild = false;\n  if ((version !== null && version !== SCHEMA_VERSION) || (features !== null && features !== wantFeatures)) {\n    // Indexing derives from presets, so a config edit rebuilding the cache must say so and\n    // name what changed -- silent rebuilds make derived indexing look like a hang or a bug.\n    if (version !== null && version !== SCHEMA_VERSION) {\n      console.error('sense: cache format changed (new sensemaking version); rebuilding the index');\n      db.close();\n      clearCache(cfg);\n      return open(cfg);\n    }\n    const changedKeys = changedSignatureKeys(features ?? '', wantFeatures);\n    // Only the tokenizer moved: frontmatter, links, sections, and embeddings are file-derived\n    // and tokenizer-independent, so they don't need re-deriving. Everything else -- a preset\n    // edit, an embed model change -- still takes the full clear/reopen below.\n    if (changedKeys.size === 1 && changedKeys.has('tokenize')) {\n      console.error('sense: config change (content tokenizer) rebuilds the text index; vectors, links, and sections are kept');\n      db.exec('DROP TABLE content');\n      tokenizeOnlyRebuild = true;\n    } else {\n      const changed = signatureDiff(features ?? '', wantFeatures);\n      console.error(`sense: config change (${changed}) rebuilds the index`);\n      db.close();\n      clearCache(cfg);\n      return open(cfg);\n    }\n  }\n\n  // Meta can lie after a crash between table creation and the signature write; the table's\n  // own DDL cannot. A mismatch here rebuilds no matter what meta says. A tokenize-only rebuild\n  // just dropped content above, so storedTokenize sees no table and this guard is skipped\n  // naturally rather than needing its own case.\n  const stored = storedTokenize(db);\n  if (stored !== null && stored !== tokenize) {\n    console.error('sense: cache was built with a different content tokenizer; rebuilding the index');\n    db.close();\n    clearCache(cfg);\n    return open(cfg);\n  }\n\n  ensureSchema(db, cfg, tokenize);\n\n  let rebuildWarnings: string[] = [];\n  if (tokenizeOnlyRebuild) {\n    rebuildWarnings = rebuildContentTable(db, cfg, cfg.baseDir);\n    setMeta(db, 'features', wantFeatures);\n  }\n\n  // 3x the largest reconcile this cache has recorded, floored at 30s and capped at 10min.\n  // Installed before reconcile() -- that call is the one that races a watcher's transaction.\n  const recordedMaxMs = Number(getMeta(db, 'reconcile_max_ms') ?? '0');\n  db.exec(`PRAGMA busy_timeout = ${Math.min(Math.max(30000, 3 * recordedMaxMs), 600_000)}`);\n\n  const { parsed, warnings } = reconcile(db, cfg, cfg.baseDir);\n\n  return { db, cfg, dbPath, parsed, warnings: [...rebuildWarnings, ...warnings] };\n}\n\n// Deletes the cache directory, and only that. The rebuild is not this function's job: the next\n// open() reconciles, which is what running any command already does, so a verb that bundled the\n// two described the half that was not its own. Manual reset for a doubted cache; the schema and\n// config-signature mismatches below reset themselves.\nexport function clearCache(cfg: ResolvedConfig): void {\n  rmSync(join(cfg.baseDir, STATE_DIR), { recursive: true, force: true });\n}\n"],"names":["mkdirSync","rmSync","join","DatabaseSync","contentTokenize","featureSignature","STATE_DIR","SenseError","activeFeatures","registerFunctions","changedSignatureKeys","rebuildContentTable","reconcile","signatureDiff","getMeta","setMeta","DB_FILENAME","SCHEMA_VERSION","DEFAULT_TOKENIZE","resolveTokenize","db","cfg","configured","undefined","literal","replace","exec","err","message","storedTokenize","row","prepare","get","m","sql","match","ensureSchema","tokenize","feature","schema","docCount","n","open","stateDir","baseDir","recursive","dbPath","version","features","wantFeatures","tokenizeOnlyRebuild","console","error","close","clearCache","changedKeys","size","has","changed","stored","rebuildWarnings","recordedMaxMs","Number","Math","min","max","parsed","warnings","force"],"mappings":"AAAA,+FAA+F;AAC/F,iGAAiG;AACjG,SAASA,SAAS,EAAEC,MAAM,QAAQ,UAAU;AAC5C,SAASC,IAAI,QAAQ,YAAY;AACjC,SAASC,YAAY,QAAQ,cAAc;AAE3C,SAASC,eAAe,EAAEC,gBAAgB,EAAEC,SAAS,QAAQ,qBAAqB;AAClF,SAASC,UAAU,QAAQ,eAAe;AAC1C,SAASC,cAAc,QAAQ,uBAAuB;AACtD,SAASC,iBAAiB,QAAQ,sBAAsB;AACxD,SAASC,oBAAoB,EAAEC,mBAAmB,EAAEC,SAAS,EAAEC,aAAa,QAAQ,iBAAiB;AACrG,SAASC,OAAO,EAAEC,OAAO,QAAQ,cAAc;AAE/C,OAAO,MAAMC,cAAc,WAAW;AACtC,sFAAsF;AACtF,iCAAiC;AACjC,OAAO,MAAMC,iBAAiB,KAAK;AAUnC,yFAAyF;AACzF,2FAA2F;AAC3F,4FAA4F;AAC5F,qCAAqC;AACrC,MAAMC,mBAAmB;AAEzB,wFAAwF;AACxF,6FAA6F;AAC7F,6FAA6F;AAC7F,6FAA6F;AAC7F,+DAA+D;AAC/D,SAASC,gBAAgBC,EAAgB,EAAEC,GAAW;IACpD,MAAMC,aAAalB,gBAAgBiB;IACnC,IAAIC,eAAeC,WAAW,OAAOL;IACrC,MAAMM,UAAUF,WAAWG,OAAO,CAAC,MAAM;IACzC,IAAI;QACFL,GAAGM,IAAI,CAAC;QACRN,GAAGM,IAAI,CAAC,CAAC,yEAAyE,EAAEF,QAAQ,EAAE,CAAC;QAC/FJ,GAAGM,IAAI,CAAC;IACV,EAAE,OAAOC,KAAK;QACZ,MAAM,IAAIpB,WAAW,kBAAkB,CAAC,kBAAkB,EAAEe,WAAW,0CAA0C,EAAE,AAACK,IAAcC,OAAO,CAAC,8FAA8F,CAAC;IAC3O;IACA,OAAOJ;AACT;AAEA,uFAAuF;AACvF,2FAA2F;AAC3F,SAASK,eAAeT,EAAgB;IACtC,MAAMU,MAAMV,GAAGW,OAAO,CAAC,CAAC,oDAAoD,CAAC,EAAEC,GAAG;IAClF,IAAI,CAACF,KAAK,OAAO;IACjB,MAAMG,IAAIH,IAAII,GAAG,CAACC,KAAK,CAAC;IACxB,OAAOF,IAAIA,CAAC,CAAC,EAAE,GAAG;AACpB;AAEA,2FAA2F;AAC3F,wFAAwF;AACxF,SAASG,aAAahB,EAAgB,EAAEC,GAAW,EAAEgB,QAAgB;IACnEjB,GAAGM,IAAI,CAAC,CAAC,oIAAoI,CAAC;IAC9I,wFAAwF;IACxF,4FAA4F;IAC5F,4FAA4F;IAC5F,2FAA2F;IAC3F,2FAA2F;IAC3F,2EAA2E;IAC3EN,GAAGM,IAAI,CAAC,CAAC,0IAA0I,EAAEW,SAAS,EAAE,CAAC;IACjK,6FAA6F;IAC7F,qFAAqF;IACrFjB,GAAGM,IAAI,CAAC,CAAC,gGAAgG,CAAC;IAC1GN,GAAGM,IAAI,CAAC;IACR,KAAK,MAAMY,WAAW9B,eAAea,KAAMiB,QAAQC,MAAM,CAACnB;IAC1D,IAAIN,QAAQM,IAAI,sBAAsB,MAAML,QAAQK,IAAI,kBAAkBH;IAC1E,IAAIH,QAAQM,IAAI,gBAAgB,MAAML,QAAQK,IAAI,YAAYf,iBAAiBgB;AACjF;AAEA,OAAO,SAASmB,SAASpB,EAAgB;IACvC,MAAMU,MAAMV,GAAGW,OAAO,CAAC,yCAAyCC,GAAG;IACnE,OAAOF,IAAIW,CAAC;AACd;AAEA,OAAO,SAASC,KAAKrB,GAAmB;QAwETP;IAvE7B,MAAM6B,WAAWzC,KAAKmB,IAAIuB,OAAO,EAAEtC;IACnCN,UAAU2C,UAAU;QAAEE,WAAW;IAAK;IACtC,MAAMC,SAAS5C,KAAKyC,UAAU3B;IAE9B,MAAMI,KAAK,IAAIjB,aAAa2C;IAC5B1B,GAAGM,IAAI,CAAC;IACR,yFAAyF;IACzF,uCAAuC;IACvCN,GAAGM,IAAI,CAAC;IACRjB,kBAAkBW,IAAIhB,gBAAgBiB,SAASE;IAE/CH,GAAGM,IAAI,CAAC;IAER,2FAA2F;IAC3F,8FAA8F;IAC9F,MAAMW,WAAWlB,gBAAgBC,IAAIC;IAErC,uFAAuF;IACvF,qGAAqG;IACrG,MAAM0B,UAAUjC,QAAQM,IAAI;IAC5B,MAAM4B,WAAWlC,QAAQM,IAAI;IAC7B,MAAM6B,eAAe5C,iBAAiBgB;IACtC,IAAI6B,sBAAsB;IAC1B,IAAI,AAACH,YAAY,QAAQA,YAAY9B,kBAAoB+B,aAAa,QAAQA,aAAaC,cAAe;QACxG,uFAAuF;QACvF,wFAAwF;QACxF,IAAIF,YAAY,QAAQA,YAAY9B,gBAAgB;YAClDkC,QAAQC,KAAK,CAAC;YACdhC,GAAGiC,KAAK;YACRC,WAAWjC;YACX,OAAOqB,KAAKrB;QACd;QACA,MAAMkC,cAAc7C,qBAAqBsC,qBAAAA,sBAAAA,WAAY,IAAIC;QACzD,0FAA0F;QAC1F,yFAAyF;QACzF,0EAA0E;QAC1E,IAAIM,YAAYC,IAAI,KAAK,KAAKD,YAAYE,GAAG,CAAC,aAAa;YACzDN,QAAQC,KAAK,CAAC;YACdhC,GAAGM,IAAI,CAAC;YACRwB,sBAAsB;QACxB,OAAO;YACL,MAAMQ,UAAU7C,cAAcmC,qBAAAA,sBAAAA,WAAY,IAAIC;YAC9CE,QAAQC,KAAK,CAAC,CAAC,sBAAsB,EAAEM,QAAQ,oBAAoB,CAAC;YACpEtC,GAAGiC,KAAK;YACRC,WAAWjC;YACX,OAAOqB,KAAKrB;QACd;IACF;IAEA,yFAAyF;IACzF,6FAA6F;IAC7F,wFAAwF;IACxF,8CAA8C;IAC9C,MAAMsC,SAAS9B,eAAeT;IAC9B,IAAIuC,WAAW,QAAQA,WAAWtB,UAAU;QAC1Cc,QAAQC,KAAK,CAAC;QACdhC,GAAGiC,KAAK;QACRC,WAAWjC;QACX,OAAOqB,KAAKrB;IACd;IAEAe,aAAahB,IAAIC,KAAKgB;IAEtB,IAAIuB,kBAA4B,EAAE;IAClC,IAAIV,qBAAqB;QACvBU,kBAAkBjD,oBAAoBS,IAAIC,KAAKA,IAAIuB,OAAO;QAC1D7B,QAAQK,IAAI,YAAY6B;IAC1B;IAEA,wFAAwF;IACxF,2FAA2F;IAC3F,MAAMY,gBAAgBC,QAAOhD,WAAAA,QAAQM,IAAI,iCAAZN,sBAAAA,WAAmC;IAChEM,GAAGM,IAAI,CAAC,CAAC,sBAAsB,EAAEqC,KAAKC,GAAG,CAACD,KAAKE,GAAG,CAAC,OAAO,IAAIJ,gBAAgB,SAAU;IAExF,MAAM,EAAEK,MAAM,EAAEC,QAAQ,EAAE,GAAGvD,UAAUQ,IAAIC,KAAKA,IAAIuB,OAAO;IAE3D,OAAO;QAAExB;QAAIC;QAAKyB;QAAQoB;QAAQC,UAAU;eAAIP;eAAoBO;SAAS;IAAC;AAChF;AAEA,+FAA+F;AAC/F,gGAAgG;AAChG,gGAAgG;AAChG,sDAAsD;AACtD,OAAO,SAASb,WAAWjC,GAAmB;IAC5CpB,OAAOC,KAAKmB,IAAIuB,OAAO,EAAEtC,YAAY;QAAEuC,WAAW;QAAMuB,OAAO;IAAK;AACtE"}
+{"version":3,"sources":["/Users/kevin/Dev/OpenSource/ai/sensemaking/src/db/open.ts"],"sourcesContent":["// The Node floor (>=22.20) is explained here and nowhere else: 22.20 is the first release with\n// both FTS5 and row-returning INSERT ... RETURNING. Raise it only for a load-bearing capability.\nimport { mkdirSync, rmSync } from 'node:fs';\nimport { join } from 'node:path';\nimport { DatabaseSync } from 'node:sqlite';\nimport type { Config, ResolvedConfig } from '../config/index.ts';\nimport { contentTokenize, featureSignature, STATE_DIR } from '../config/index.ts';\nimport { SenseError } from '../errors.ts';\nimport { activeFeatures } from '../features/index.ts';\nimport { registerFunctions } from '../sql-functions.ts';\nimport { changedSignatureKeys, embedIdentityAdopted, rebuildContentTable, reconcile, signatureDiff } from './reconcile.ts';\nimport { getMeta, setMeta } from './shared.ts';\n\nexport const DB_FILENAME = 'cache.db';\n// Cache shape version, independent of the config's own `version`. Bumping it rebuilds\n// existing trees on first query.\nexport const SCHEMA_VERSION = '17';\n\nexport interface OpenResult {\n  db: DatabaseSync;\n  cfg: ResolvedConfig;\n  dbPath: string;\n  parsed: number;\n  warnings: string[];\n}\n\n// Stemming is English-only, but the segmentation underneath it is what decides coverage:\n// unicode61 splits on spaces, so a language written without them (Chinese, Japanese, Thai)\n// indexes a whole run as one token and word search finds nothing. `content.tokenize` is how\n// such a tree picks trigram instead.\nconst DEFAULT_TOKENIZE = 'porter unicode61';\n\n// FTS5 takes its tokenizer as a string literal inside DDL, where nothing can bind, so a\n// configured value has to be concatenated. Probing a throwaway table is what makes that safe\n// and is also the whole validation: anything the linked SQLite accepts passes, anything else\n// fails here with SQLite's own message rather than against the real table. It means no table\n// of which version added which tokenizer has to be maintained.\nfunction resolveTokenize(db: DatabaseSync, cfg: Config): string {\n  const configured = contentTokenize(cfg);\n  if (configured === undefined) return DEFAULT_TOKENIZE;\n  const literal = configured.replace(/'/g, \"''\");\n  try {\n    db.exec('DROP TABLE IF EXISTS temp.sense_tokenize_probe');\n    db.exec(`CREATE VIRTUAL TABLE temp.sense_tokenize_probe USING fts5(x, tokenize = '${literal}')`);\n    db.exec('DROP TABLE IF EXISTS temp.sense_tokenize_probe');\n  } catch (err) {\n    throw new SenseError('CONFIG_INVALID', `content.tokenize \"${configured}\" is not a tokenizer this SQLite accepts (${(err as Error).message}); the built-in choices are unicode61, ascii, porter, and trigram, each with their own options`);\n  }\n  return literal;\n}\n\n// The tokenizer the content table was actually built with, from its own DDL -- the one\n// record that cannot desynchronize from the table. NULL when the table does not exist yet.\nfunction storedTokenize(db: DatabaseSync): string | null {\n  const row = db.prepare(`SELECT sql FROM sqlite_master WHERE name = 'content'`).get() as { sql: string } | undefined;\n  if (!row) return null;\n  const m = row.sql.match(/tokenize = '((?:[^']|'')*)'/);\n  return m ? m[1] : null;\n}\n\n// Content is a separate table (not a column on frontmatter) so `SELECT * FROM frontmatter`\n// can't dump file text into context. Features add their own tables after the core ones.\nfunction ensureSchema(db: DatabaseSync, cfg: Config, tokenize: string): void {\n  db.exec(`CREATE TABLE IF NOT EXISTS frontmatter (\"path\" TEXT PRIMARY KEY, \"_mtime\" REAL, \"_ctime\" REAL, \"_size\" INTEGER, \"_parse_error\" TEXT)`);\n  // IF NOT EXISTS is safe against a tokenizer change: open() compares the table's own DDL\n  // against the resolved tokenizer before this runs, so a stale table is already gone by now.\n  // The three `_seg` sidecars are appended after path, never inserted: bm25(content, ...) and\n  // snippet(content, 2, ...) are documented against the first three columns and keep working\n  // (FTS5 defaults the weights it was not given). Each carries its field's exploded unspaced\n  // runs for text that needs it, and an empty string for text that does not.\n  db.exec(`CREATE VIRTUAL TABLE IF NOT EXISTS content USING fts5(title, summary, text, path UNINDEXED, title_seg, summary_seg, text_seg, tokenize = '${tokenize}')`);\n  // Coverage, not ownership: a path can appear under several presets. path leads the PK so the\n  // per-doc delete is an index hit -- keyed the other way, cold builds went quadratic.\n  db.exec(`CREATE TABLE IF NOT EXISTS preset_files (\"path\" TEXT, preset TEXT, PRIMARY KEY (\"path\", preset))`);\n  db.exec('CREATE INDEX IF NOT EXISTS preset_files_preset ON preset_files(preset)');\n  for (const feature of activeFeatures(cfg)) feature.schema(db);\n  if (getMeta(db, 'schema_version') === null) setMeta(db, 'schema_version', SCHEMA_VERSION);\n  if (getMeta(db, 'features') === null) setMeta(db, 'features', featureSignature(cfg));\n}\n\nexport function docCount(db: DatabaseSync): number {\n  const row = db.prepare('SELECT COUNT(*) AS n FROM frontmatter').get() as { n: number };\n  return row.n;\n}\n\nexport function open(cfg: ResolvedConfig): OpenResult {\n  const stateDir = join(cfg.baseDir, STATE_DIR);\n  mkdirSync(stateDir, { recursive: true });\n  const dbPath = join(stateDir, DB_FILENAME);\n\n  const db = new DatabaseSync(dbPath);\n  db.exec('PRAGMA journal_mode = WAL');\n  // Covers a concurrent watcher's bulk reconcile (~5s for 500 files at 26k notes). A query\n  // that outwaits it still fails loudly.\n  db.exec('PRAGMA busy_timeout = 30000');\n  registerFunctions(db, contentTokenize(cfg) === undefined);\n\n  db.exec('CREATE TABLE IF NOT EXISTS meta (key TEXT PRIMARY KEY, value TEXT)');\n\n  // Before the rebuild branch below, never after: that branch deletes the cache, so a typo'd\n  // tokenizer validated later would cost a full re-index (and re-embed) to reach its own error.\n  const tokenize = resolveTokenize(db, cfg);\n\n  // Schema-version or feature-set mismatch: reconcile only reparses changed files, so an\n  // old cache can't be patched incrementally -- rebuild instead (cheap: nothing expensive lives here).\n  const version = getMeta(db, 'schema_version');\n  const features = getMeta(db, 'features');\n  const wantFeatures = featureSignature(cfg);\n  let tokenizeOnlyRebuild = false;\n  if ((version !== null && version !== SCHEMA_VERSION) || (features !== null && features !== wantFeatures)) {\n    // Indexing derives from presets, so a config edit rebuilding the cache must say so and\n    // name what changed -- silent rebuilds make derived indexing look like a hang or a bug.\n    if (version !== null && version !== SCHEMA_VERSION) {\n      console.error('sense: cache format changed (new sensemaking version); rebuilding the index');\n      db.close();\n      clearCache(cfg);\n      return open(cfg);\n    }\n    const changedKeys = changedSignatureKeys(features ?? '', wantFeatures);\n    // Only the tokenizer moved: frontmatter, links, sections, and embeddings are file-derived\n    // and tokenizer-independent, so they don't need re-deriving. Everything else -- a preset\n    // edit, an embed model change -- still takes the full clear/reopen below.\n    if (changedKeys.size === 1 && changedKeys.has('tokenize')) {\n      console.error('sense: config change (content tokenizer) rebuilds the text index; vectors, links, and sections are kept');\n      db.exec('DROP TABLE content');\n      tokenizeOnlyRebuild = true;\n    } else if (changedKeys.size === 1 && changedKeys.has('embed') && embedIdentityAdopted(features ?? '', wantFeatures)) {\n      // First sight of a resolved weight identity: the model itself hasn't changed, so\n      // adopt it into meta with no rebuild and no re-embed, mirroring the tokenize precedent.\n      console.error(\"sense: recorded the embedding model's resolved identity; vectors are unaffected\");\n      setMeta(db, 'features', wantFeatures);\n    } else {\n      const changed = signatureDiff(features ?? '', wantFeatures);\n      console.error(`sense: config change (${changed}) rebuilds the index`);\n      db.close();\n      clearCache(cfg);\n      return open(cfg);\n    }\n  }\n\n  // Meta can lie after a crash between table creation and the signature write; the table's\n  // own DDL cannot. A mismatch here rebuilds no matter what meta says. A tokenize-only rebuild\n  // just dropped content above, so storedTokenize sees no table and this guard is skipped\n  // naturally rather than needing its own case.\n  const stored = storedTokenize(db);\n  if (stored !== null && stored !== tokenize) {\n    console.error('sense: cache was built with a different content tokenizer; rebuilding the index');\n    db.close();\n    clearCache(cfg);\n    return open(cfg);\n  }\n\n  ensureSchema(db, cfg, tokenize);\n\n  let rebuildWarnings: string[] = [];\n  if (tokenizeOnlyRebuild) {\n    rebuildWarnings = rebuildContentTable(db, cfg, cfg.baseDir);\n    setMeta(db, 'features', wantFeatures);\n  }\n\n  // 3x the largest reconcile this cache has recorded, floored at 30s and capped at 10min.\n  // Installed before reconcile() -- that call is the one that races a watcher's transaction.\n  const recordedMaxMs = Number(getMeta(db, 'reconcile_max_ms') ?? '0');\n  db.exec(`PRAGMA busy_timeout = ${Math.min(Math.max(30000, 3 * recordedMaxMs), 600_000)}`);\n\n  const { parsed, warnings } = reconcile(db, cfg, cfg.baseDir);\n\n  return { db, cfg, dbPath, parsed, warnings: [...rebuildWarnings, ...warnings] };\n}\n\n// Deletes the cache directory, and only that. The rebuild is not this function's job: the next\n// open() reconciles, which is what running any command already does, so a verb that bundled the\n// two described the half that was not its own. Manual reset for a doubted cache; the schema and\n// config-signature mismatches below reset themselves.\nexport function clearCache(cfg: ResolvedConfig): void {\n  rmSync(join(cfg.baseDir, STATE_DIR), { recursive: true, force: true });\n}\n"],"names":["mkdirSync","rmSync","join","DatabaseSync","contentTokenize","featureSignature","STATE_DIR","SenseError","activeFeatures","registerFunctions","changedSignatureKeys","embedIdentityAdopted","rebuildContentTable","reconcile","signatureDiff","getMeta","setMeta","DB_FILENAME","SCHEMA_VERSION","DEFAULT_TOKENIZE","resolveTokenize","db","cfg","configured","undefined","literal","replace","exec","err","message","storedTokenize","row","prepare","get","m","sql","match","ensureSchema","tokenize","feature","schema","docCount","n","open","stateDir","baseDir","recursive","dbPath","version","features","wantFeatures","tokenizeOnlyRebuild","console","error","close","clearCache","changedKeys","size","has","changed","stored","rebuildWarnings","recordedMaxMs","Number","Math","min","max","parsed","warnings","force"],"mappings":"AAAA,+FAA+F;AAC/F,iGAAiG;AACjG,SAASA,SAAS,EAAEC,MAAM,QAAQ,UAAU;AAC5C,SAASC,IAAI,QAAQ,YAAY;AACjC,SAASC,YAAY,QAAQ,cAAc;AAE3C,SAASC,eAAe,EAAEC,gBAAgB,EAAEC,SAAS,QAAQ,qBAAqB;AAClF,SAASC,UAAU,QAAQ,eAAe;AAC1C,SAASC,cAAc,QAAQ,uBAAuB;AACtD,SAASC,iBAAiB,QAAQ,sBAAsB;AACxD,SAASC,oBAAoB,EAAEC,oBAAoB,EAAEC,mBAAmB,EAAEC,SAAS,EAAEC,aAAa,QAAQ,iBAAiB;AAC3H,SAASC,OAAO,EAAEC,OAAO,QAAQ,cAAc;AAE/C,OAAO,MAAMC,cAAc,WAAW;AACtC,sFAAsF;AACtF,iCAAiC;AACjC,OAAO,MAAMC,iBAAiB,KAAK;AAUnC,yFAAyF;AACzF,2FAA2F;AAC3F,4FAA4F;AAC5F,qCAAqC;AACrC,MAAMC,mBAAmB;AAEzB,wFAAwF;AACxF,6FAA6F;AAC7F,6FAA6F;AAC7F,6FAA6F;AAC7F,+DAA+D;AAC/D,SAASC,gBAAgBC,EAAgB,EAAEC,GAAW;IACpD,MAAMC,aAAanB,gBAAgBkB;IACnC,IAAIC,eAAeC,WAAW,OAAOL;IACrC,MAAMM,UAAUF,WAAWG,OAAO,CAAC,MAAM;IACzC,IAAI;QACFL,GAAGM,IAAI,CAAC;QACRN,GAAGM,IAAI,CAAC,CAAC,yEAAyE,EAAEF,QAAQ,EAAE,CAAC;QAC/FJ,GAAGM,IAAI,CAAC;IACV,EAAE,OAAOC,KAAK;QACZ,MAAM,IAAIrB,WAAW,kBAAkB,CAAC,kBAAkB,EAAEgB,WAAW,0CAA0C,EAAE,AAACK,IAAcC,OAAO,CAAC,8FAA8F,CAAC;IAC3O;IACA,OAAOJ;AACT;AAEA,uFAAuF;AACvF,2FAA2F;AAC3F,SAASK,eAAeT,EAAgB;IACtC,MAAMU,MAAMV,GAAGW,OAAO,CAAC,CAAC,oDAAoD,CAAC,EAAEC,GAAG;IAClF,IAAI,CAACF,KAAK,OAAO;IACjB,MAAMG,IAAIH,IAAII,GAAG,CAACC,KAAK,CAAC;IACxB,OAAOF,IAAIA,CAAC,CAAC,EAAE,GAAG;AACpB;AAEA,2FAA2F;AAC3F,wFAAwF;AACxF,SAASG,aAAahB,EAAgB,EAAEC,GAAW,EAAEgB,QAAgB;IACnEjB,GAAGM,IAAI,CAAC,CAAC,oIAAoI,CAAC;IAC9I,wFAAwF;IACxF,4FAA4F;IAC5F,4FAA4F;IAC5F,2FAA2F;IAC3F,2FAA2F;IAC3F,2EAA2E;IAC3EN,GAAGM,IAAI,CAAC,CAAC,0IAA0I,EAAEW,SAAS,EAAE,CAAC;IACjK,6FAA6F;IAC7F,qFAAqF;IACrFjB,GAAGM,IAAI,CAAC,CAAC,gGAAgG,CAAC;IAC1GN,GAAGM,IAAI,CAAC;IACR,KAAK,MAAMY,WAAW/B,eAAec,KAAMiB,QAAQC,MAAM,CAACnB;IAC1D,IAAIN,QAAQM,IAAI,sBAAsB,MAAML,QAAQK,IAAI,kBAAkBH;IAC1E,IAAIH,QAAQM,IAAI,gBAAgB,MAAML,QAAQK,IAAI,YAAYhB,iBAAiBiB;AACjF;AAEA,OAAO,SAASmB,SAASpB,EAAgB;IACvC,MAAMU,MAAMV,GAAGW,OAAO,CAAC,yCAAyCC,GAAG;IACnE,OAAOF,IAAIW,CAAC;AACd;AAEA,OAAO,SAASC,KAAKrB,GAAmB;QA6ETP;IA5E7B,MAAM6B,WAAW1C,KAAKoB,IAAIuB,OAAO,EAAEvC;IACnCN,UAAU4C,UAAU;QAAEE,WAAW;IAAK;IACtC,MAAMC,SAAS7C,KAAK0C,UAAU3B;IAE9B,MAAMI,KAAK,IAAIlB,aAAa4C;IAC5B1B,GAAGM,IAAI,CAAC;IACR,yFAAyF;IACzF,uCAAuC;IACvCN,GAAGM,IAAI,CAAC;IACRlB,kBAAkBY,IAAIjB,gBAAgBkB,SAASE;IAE/CH,GAAGM,IAAI,CAAC;IAER,2FAA2F;IAC3F,8FAA8F;IAC9F,MAAMW,WAAWlB,gBAAgBC,IAAIC;IAErC,uFAAuF;IACvF,qGAAqG;IACrG,MAAM0B,UAAUjC,QAAQM,IAAI;IAC5B,MAAM4B,WAAWlC,QAAQM,IAAI;IAC7B,MAAM6B,eAAe7C,iBAAiBiB;IACtC,IAAI6B,sBAAsB;IAC1B,IAAI,AAACH,YAAY,QAAQA,YAAY9B,kBAAoB+B,aAAa,QAAQA,aAAaC,cAAe;QACxG,uFAAuF;QACvF,wFAAwF;QACxF,IAAIF,YAAY,QAAQA,YAAY9B,gBAAgB;YAClDkC,QAAQC,KAAK,CAAC;YACdhC,GAAGiC,KAAK;YACRC,WAAWjC;YACX,OAAOqB,KAAKrB;QACd;QACA,MAAMkC,cAAc9C,qBAAqBuC,qBAAAA,sBAAAA,WAAY,IAAIC;QACzD,0FAA0F;QAC1F,yFAAyF;QACzF,0EAA0E;QAC1E,IAAIM,YAAYC,IAAI,KAAK,KAAKD,YAAYE,GAAG,CAAC,aAAa;YACzDN,QAAQC,KAAK,CAAC;YACdhC,GAAGM,IAAI,CAAC;YACRwB,sBAAsB;QACxB,OAAO,IAAIK,YAAYC,IAAI,KAAK,KAAKD,YAAYE,GAAG,CAAC,YAAY/C,qBAAqBsC,qBAAAA,sBAAAA,WAAY,IAAIC,eAAe;YACnH,iFAAiF;YACjF,wFAAwF;YACxFE,QAAQC,KAAK,CAAC;YACdrC,QAAQK,IAAI,YAAY6B;QAC1B,OAAO;YACL,MAAMS,UAAU7C,cAAcmC,qBAAAA,sBAAAA,WAAY,IAAIC;YAC9CE,QAAQC,KAAK,CAAC,CAAC,sBAAsB,EAAEM,QAAQ,oBAAoB,CAAC;YACpEtC,GAAGiC,KAAK;YACRC,WAAWjC;YACX,OAAOqB,KAAKrB;QACd;IACF;IAEA,yFAAyF;IACzF,6FAA6F;IAC7F,wFAAwF;IACxF,8CAA8C;IAC9C,MAAMsC,SAAS9B,eAAeT;IAC9B,IAAIuC,WAAW,QAAQA,WAAWtB,UAAU;QAC1Cc,QAAQC,KAAK,CAAC;QACdhC,GAAGiC,KAAK;QACRC,WAAWjC;QACX,OAAOqB,KAAKrB;IACd;IAEAe,aAAahB,IAAIC,KAAKgB;IAEtB,IAAIuB,kBAA4B,EAAE;IAClC,IAAIV,qBAAqB;QACvBU,kBAAkBjD,oBAAoBS,IAAIC,KAAKA,IAAIuB,OAAO;QAC1D7B,QAAQK,IAAI,YAAY6B;IAC1B;IAEA,wFAAwF;IACxF,2FAA2F;IAC3F,MAAMY,gBAAgBC,QAAOhD,WAAAA,QAAQM,IAAI,iCAAZN,sBAAAA,WAAmC;IAChEM,GAAGM,IAAI,CAAC,CAAC,sBAAsB,EAAEqC,KAAKC,GAAG,CAACD,KAAKE,GAAG,CAAC,OAAO,IAAIJ,gBAAgB,SAAU;IAExF,MAAM,EAAEK,MAAM,EAAEC,QAAQ,EAAE,GAAGvD,UAAUQ,IAAIC,KAAKA,IAAIuB,OAAO;IAE3D,OAAO;QAAExB;QAAIC;QAAKyB;QAAQoB;QAAQC,UAAU;eAAIP;eAAoBO;SAAS;IAAC;AAChF;AAEA,+FAA+F;AAC/F,gGAAgG;AAChG,gGAAgG;AAChG,sDAAsD;AACtD,OAAO,SAASb,WAAWjC,GAAmB;IAC5CrB,OAAOC,KAAKoB,IAAIuB,OAAO,EAAEvC,YAAY;QAAEwC,WAAW;QAAMuB,OAAO;IAAK;AACtE"}

package/dist/esm/db/reconcile.d.ts

@@ -5,5 +5,6 @@ export declare function reconcile(db: DatabaseSync, cfg: Config, baseDir: string
     warnings: string[];
 };
 export declare function changedSignatureKeys(before: string, after: string): Set<string>;
+export declare function embedIdentityAdopted(before: string, after: string): boolean;
 export declare function signatureDiff(before: string, after: string): string;
 export declare function rebuildContentTable(db: DatabaseSync, cfg: Config, baseDir: string): string[];

package/dist/esm/db/reconcile.js

@@ -178,6 +178,16 @@ export function changedSignatureKeys(before, after) {
     for (const key of a.keys())if (!b.has(key)) changed.add(key);
     return changed;
 }
+// Whether the embed segment only gained its resolved weight identity: same provider and
+// model, no identity recorded before, one now -- adopted into meta without a rebuild.
+export function embedIdentityAdopted(before, after) {
+    const embedPart = (sig)=>sig.split('|').find((p)=>p.startsWith('embed:'));
+    const b = embedPart(before);
+    const a = embedPart(after);
+    if (b === undefined || a === undefined) return false;
+    const at = a.indexOf('@');
+    return b.indexOf('@') === -1 && at !== -1 && a.slice(0, at) === b;
+}
 // Names what moved, for the rebuild notice.
 export function signatureDiff(before, after) {
     const changed = changedSignatureKeys(before, after);

package/dist/esm/db/reconcile.js.map

@@ -1 +1 @@
-{"version":3,"sources":["/Users/kevin/Dev/OpenSource/ai/sensemaking/src/db/reconcile.ts"],"sourcesContent":["import type { DatabaseSync, StatementSync } from 'node:sqlite';\nimport type { Config } from '../config/index.ts';\nimport { contentTokenize } from '../config/index.ts';\nimport { SenseError } from '../errors.ts';\nimport { activeFeatures } from '../features/index.ts';\nimport type { ReconcileDelta } from '../features/types.ts';\nimport { progress } from '../progress.ts';\nimport type { ParsedDoc } from '../scan.ts';\nimport { listFiles, parseFile, RESERVED_COLUMNS } from '../scan.ts';\nimport { segmentField } from '../segment.ts';\nimport { getColumns, getMeta, quoteIdent, setMeta } from './shared.ts';\n\n// Feature-owned columns (`_rank`) must stay out of the upsert: a reparse would null the last\n// computed value on every touch, not just the reconciles that recompute it.\nconst CORE_FRONTMATTER_COLUMNS = new Set(['path', '_mtime', '_ctime', '_size', '_parse_error']);\n\n// SQLite's compile-time SQLITE_MAX_COLUMN, default 2000 (https://www.sqlite.org/limits.html).\nconst MAX_FRONTMATTER_COLUMNS = 2000;\n\n// Shared by reconcile() and the tokenize-only rebuild in open(), so both prepare the same\n// literal instead of two copies drifting apart.\nconst INSERT_CONTENT_SQL = `INSERT INTO content (rowid, title, summary, text, \"path\", title_seg, summary_seg, text_seg) VALUES ((SELECT rowid FROM frontmatter WHERE \"path\" = ?), ?, ?, ?, ?, ?, ?, ?)`;\n\n// Shared by reconcile() and the tokenize-only rebuild in open(), so content population is\n// defined once. Assumes the frontmatter row for doc.relPath already exists (rowid lookup).\nfunction insertContentRow(insertBody: StatementSync, doc: ParsedDoc, segmenting: boolean): void {\n  insertBody.run(doc.relPath, doc.search.title, doc.search.summary, doc.search.text, doc.relPath, segmenting ? segmentField(doc.search.title) : '', segmenting ? segmentField(doc.search.summary) : '', segmenting ? segmentField(doc.search.text) : '');\n}\n\nexport function reconcile(db: DatabaseSync, cfg: Config, baseDir: string): { parsed: number; warnings: string[] } {\n  const files = listFiles(cfg, baseDir);\n  const currentSet = new Set(files.map((f) => f.relPath));\n\n  const existingRows = db.prepare(`SELECT \"path\", \"_mtime\", \"_size\" FROM frontmatter`).all() as Array<{\n    path: string;\n    _mtime: number;\n    _size: number;\n  }>;\n  const existing = new Map(existingRows.map((r) => [r.path, r]));\n  const vanished = existingRows.filter((r) => !currentSet.has(r.path)).map((r) => r.path);\n\n  const toReparse = files.filter((f) => {\n    const row = existing.get(f.relPath);\n    return !row || row._mtime !== f.mtimeMs || row._size !== f.size;\n  });\n\n  if (vanished.length === 0 && toReparse.length === 0) return { parsed: 0, warnings: [] };\n\n  const features = activeFeatures(cfg);\n  const seenColumns = getColumns(db);\n  const newColumns: string[] = [];\n  const parsedDocs: ParsedDoc[] = [];\n  const warnings: string[] = [];\n\n  // Bulk reparses (a sync, a cold build) are the long silences a query can hit; short\n  // reconciles stay silent (progress() has a threshold).\n  const report = progress('reparsing files', toReparse.length);\n  let parsedCount = 0;\n  for (const file of toReparse) {\n    // A doc only gets extract/store from features that apply to it (currently: embed, via\n    // FileStat.embed -- true iff the config names an embedding model).\n    const fileFeatures = features.filter((feature) => !feature.enabledForFile || feature.enabledForFile(cfg, file));\n    const { doc, warnings: fileWarnings } = parseFile(file, fileFeatures);\n    report.tick(++parsedCount);\n    warnings.push(...fileWarnings);\n    for (const key of Object.keys(doc.data)) {\n      if (!seenColumns.has(key)) {\n        seenColumns.add(key);\n        newColumns.push(key);\n      }\n    }\n    parsedDocs.push(doc);\n  }\n  report.finish();\n\n  const allColumns = [...seenColumns];\n  // Fence before ALTERing: SQLite's own failure past this point is a raw\n  // \"too many columns on sqlite_altertab_frontmatter\" with no indication of the boundary or the levers.\n  if (allColumns.length > MAX_FRONTMATTER_COLUMNS) {\n    throw new SenseError(\n      'COLUMN_LIMIT',\n      `frontmatter would need ${allColumns.length} columns, crossing SQLite's compile-time SQLITE_MAX_COLUMN limit (default ${MAX_FRONTMATTER_COLUMNS}; see https://www.sqlite.org/limits.html). Narrow the presets' include globs so fewer/other files are indexed, or fix whatever is generating unbounded frontmatter keys.`\n    );\n  }\n  // Columns the frontmatter upsert actually writes: core + parsed frontmatter keys, never a\n  // feature-owned reserved column (see CORE_FRONTMATTER_COLUMNS above).\n  const writableColumns = allColumns.filter((c) => CORE_FRONTMATTER_COLUMNS.has(c) || !RESERVED_COLUMNS.has(c));\n  // ON CONFLICT UPDATE (not OR REPLACE) keeps the row's rowid stable across reparses --\n  // content rows are coupled to that rowid below.\n  const insertSql = `INSERT INTO frontmatter (${writableColumns.map(quoteIdent).join(', ')}) VALUES (${writableColumns.map(() => '?').join(', ')}) ON CONFLICT(\"path\") DO UPDATE SET ${writableColumns\n    .filter((c) => c !== 'path')\n    .map((c) => `${quoteIdent(c)} = excluded.${quoteIdent(c)}`)\n    .join(', ')}`;\n\n  const added = toReparse.filter((f) => !existing.has(f.relPath)).map((f) => f.relPath);\n  const delta: ReconcileDelta = { files, reparsed: parsedDocs.map((d) => d.relPath), added, vanished };\n\n  const txStart = Date.now();\n  db.exec('BEGIN');\n  try {\n    for (const col of newColumns) db.exec(`ALTER TABLE frontmatter ADD COLUMN ${quoteIdent(col)}`);\n    // FTS5 has no upsert, so delete-before-insert into `content`; coupled to the\n    // frontmatter rowid (indexed via its PRIMARY KEY) instead of the UNINDEXED `path`\n    // column, which a per-row DELETE would otherwise scan the whole table to find.\n    const delBody = db.prepare(`DELETE FROM content WHERE rowid = (SELECT rowid FROM frontmatter WHERE \"path\" = ?)`);\n    const delPresetFiles = db.prepare(`DELETE FROM preset_files WHERE \"path\" = ?`);\n    const insertPresetFile = db.prepare(`INSERT INTO preset_files (\"path\", preset) VALUES (?, ?)`);\n    if (vanished.length > 0) {\n      const del = db.prepare(`DELETE FROM frontmatter WHERE \"path\" = ?`);\n      for (const path of vanished) {\n        // content delete must run first: it looks up the frontmatter rowid by path, which\n        // the frontmatter delete below would otherwise have already removed.\n        delBody.run(path);\n        del.run(path);\n        delPresetFiles.run(path);\n        for (const feature of features) feature.remove?.(db, path, delta);\n      }\n    }\n    if (parsedDocs.length > 0) {\n      const insert = db.prepare(insertSql);\n      const insertBody = db.prepare(INSERT_CONTENT_SQL);\n      // A non-default tokenizer means the tree has chosen its own scheme; a phrase query over\n      // grapheme runs would be nonsense against trigram, so the sidecars stay empty.\n      const segmenting = contentTokenize(cfg) === undefined;\n      for (const doc of parsedDocs) {\n        const values = writableColumns.map((col) => {\n          if (col === 'path') return doc.relPath;\n          if (col === '_mtime') return doc.mtimeMs;\n          if (col === '_ctime') return doc.ctimeMs;\n          if (col === '_size') return doc.size;\n          // Written per parse, unlike _rank, which a feature pass owns and the upsert skips.\n          if (col === '_parse_error') return doc.parseError;\n          return doc.data[col] ?? null;\n        });\n        // Frontmatter upsert first: content's rowid lookup below depends on this row existing.\n        insert.run(...values);\n        // Only for docs that have rows: an unconditional delete made cold crawls quadratic.\n        if (existing.has(doc.relPath)) {\n          delBody.run(doc.relPath);\n          for (const feature of features) feature.remove?.(db, doc.relPath, delta);\n        }\n        insertContentRow(insertBody, doc, segmenting);\n        // A preset edit forces a full rebuild, so an unchanged doc's coverage is already\n        // correct; new docs have nothing to clear, which keeps cold builds linear.\n        if (existing.has(doc.relPath)) delPresetFiles.run(doc.relPath);\n        for (const presetName of doc.presets) insertPresetFile.run(doc.relPath, presetName);\n        for (const feature of features) feature.store?.(db, doc.relPath, doc.extracted[feature.name], delta);\n      }\n    }\n    for (const feature of features) feature.afterReconcile?.(db, delta);\n    db.exec('COMMIT');\n  } catch (err) {\n    db.exec('ROLLBACK');\n    throw err;\n  }\n\n  // Reconcile's own write-transaction duration, for open()'s derived busy_timeout (F):\n  // keep the observed max so a big watcher reconcile's lock hold is what the next open bounds its wait against.\n  const durationMs = Date.now() - txStart;\n  const prevRaw = getMeta(db, 'reconcile_max_ms');\n  // -1, not 0, so a genuinely 0ms first reconcile (sub-millisecond, common on a tiny tree)\n  // still gets recorded instead of losing to the \"nothing recorded yet\" default.\n  const prevMax = prevRaw === null ? -1 : Number(prevRaw);\n  if (durationMs > prevMax) setMeta(db, 'reconcile_max_ms', String(durationMs));\n\n  return { parsed: parsedDocs.length, warnings };\n}\n\n// Segment keys that moved between two feature signatures (see config.featureSignature's\n// format: global features, embed provider, tokenize, then one segment per preset).\nexport function changedSignatureKeys(before: string, after: string): Set<string> {\n  const keyOf = (part: string) => (part.startsWith('preset:') ? part.split(':').slice(0, 2).join(':') : part.split(':')[0]);\n  const parse = (sig: string) => new Map(sig.split('|').map((part) => [keyOf(part), part]));\n  const a = parse(before);\n  const b = parse(after);\n  const changed = new Set<string>();\n  for (const [key, val] of b) if (a.get(key) !== val) changed.add(key);\n  for (const key of a.keys()) if (!b.has(key)) changed.add(key);\n  return changed;\n}\n\n// Names what moved, for the rebuild notice.\nexport function signatureDiff(before: string, after: string): string {\n  const changed = changedSignatureKeys(before, after);\n  const label = (key: string) => (key === 'embed' ? 'embed settings' : key === 'tokenize' ? 'content tokenizer' : key.startsWith('preset:') ? `preset \"${key.slice(7)}\"` : 'features');\n  return changed.size === 0 ? 'features' : [...changed].map(label).join(', ');\n}\n\n// The tokenize-only rebuild in open(): content is dropped and repopulated from files already\n// listed in frontmatter, which itself is untouched. Frontmatter, links, sections, and\n// embeddings are file-derived and tokenizer-independent, so they survive. No feature extractors\n// run here -- doc.search (title/summary/text) is all content population needs. Returns\n// parseFile's per-file warnings (e.g. a bad date) so open() can surface them -- mtimes are\n// untouched, so reconcile() never reparses these files and would otherwise never emit them again.\nexport function rebuildContentTable(db: DatabaseSync, cfg: Config, baseDir: string): string[] {\n  const known = new Set((db.prepare('SELECT \"path\" FROM frontmatter').all() as Array<{ path: string }>).map((r) => r.path));\n  const files = listFiles(cfg, baseDir).filter((f) => known.has(f.relPath));\n  const segmenting = contentTokenize(cfg) === undefined;\n  const insertBody = db.prepare(INSERT_CONTENT_SQL);\n  const warnings: string[] = [];\n  db.exec('BEGIN');\n  try {\n    for (const file of files) {\n      const { doc, warnings: fileWarnings } = parseFile(file);\n      warnings.push(...fileWarnings);\n      insertContentRow(insertBody, doc, segmenting);\n    }\n    db.exec('COMMIT');\n  } catch (err) {\n    db.exec('ROLLBACK');\n    throw err;\n  }\n  return warnings;\n}\n"],"names":["contentTokenize","SenseError","activeFeatures","progress","listFiles","parseFile","RESERVED_COLUMNS","segmentField","getColumns","getMeta","quoteIdent","setMeta","CORE_FRONTMATTER_COLUMNS","Set","MAX_FRONTMATTER_COLUMNS","INSERT_CONTENT_SQL","insertContentRow","insertBody","doc","segmenting","run","relPath","search","title","summary","text","reconcile","db","cfg","baseDir","files","currentSet","map","f","existingRows","prepare","all","existing","Map","r","path","vanished","filter","has","toReparse","row","get","_mtime","mtimeMs","_size","size","length","parsed","warnings","features","seenColumns","newColumns","parsedDocs","report","parsedCount","file","fileFeatures","feature","enabledForFile","fileWarnings","tick","push","key","Object","keys","data","add","finish","allColumns","writableColumns","c","insertSql","join","added","delta","reparsed","d","txStart","Date","now","exec","col","delBody","delPresetFiles","insertPresetFile","del","remove","insert","undefined","values","ctimeMs","parseError","presetName","presets","store","extracted","name","afterReconcile","err","durationMs","prevRaw","prevMax","Number","String","changedSignatureKeys","before","after","keyOf","part","startsWith","split","slice","parse","sig","a","b","changed","val","signatureDiff","label","rebuildContentTable","known"],"mappings":"AAEA,SAASA,eAAe,QAAQ,qBAAqB;AACrD,SAASC,UAAU,QAAQ,eAAe;AAC1C,SAASC,cAAc,QAAQ,uBAAuB;AAEtD,SAASC,QAAQ,QAAQ,iBAAiB;AAE1C,SAASC,SAAS,EAAEC,SAAS,EAAEC,gBAAgB,QAAQ,aAAa;AACpE,SAASC,YAAY,QAAQ,gBAAgB;AAC7C,SAASC,UAAU,EAAEC,OAAO,EAAEC,UAAU,EAAEC,OAAO,QAAQ,cAAc;AAEvE,6FAA6F;AAC7F,4EAA4E;AAC5E,MAAMC,2BAA2B,IAAIC,IAAI;IAAC;IAAQ;IAAU;IAAU;IAAS;CAAe;AAE9F,8FAA8F;AAC9F,MAAMC,0BAA0B;AAEhC,0FAA0F;AAC1F,gDAAgD;AAChD,MAAMC,qBAAqB,CAAC,0KAA0K,CAAC;AAEvM,0FAA0F;AAC1F,2FAA2F;AAC3F,SAASC,iBAAiBC,UAAyB,EAAEC,GAAc,EAAEC,UAAmB;IACtFF,WAAWG,GAAG,CAACF,IAAIG,OAAO,EAAEH,IAAII,MAAM,CAACC,KAAK,EAAEL,IAAII,MAAM,CAACE,OAAO,EAAEN,IAAII,MAAM,CAACG,IAAI,EAAEP,IAAIG,OAAO,EAAEF,aAAaZ,aAAaW,IAAII,MAAM,CAACC,KAAK,IAAI,IAAIJ,aAAaZ,aAAaW,IAAII,MAAM,CAACE,OAAO,IAAI,IAAIL,aAAaZ,aAAaW,IAAII,MAAM,CAACG,IAAI,IAAI;AACrP;AAEA,OAAO,SAASC,UAAUC,EAAgB,EAAEC,GAAW,EAAEC,OAAe;IACtE,MAAMC,QAAQ1B,UAAUwB,KAAKC;IAC7B,MAAME,aAAa,IAAIlB,IAAIiB,MAAME,GAAG,CAAC,CAACC,IAAMA,EAAEZ,OAAO;IAErD,MAAMa,eAAeP,GAAGQ,OAAO,CAAC,CAAC,iDAAiD,CAAC,EAAEC,GAAG;IAKxF,MAAMC,WAAW,IAAIC,IAAIJ,aAAaF,GAAG,CAAC,CAACO,IAAM;YAACA,EAAEC,IAAI;YAAED;SAAE;IAC5D,MAAME,WAAWP,aAAaQ,MAAM,CAAC,CAACH,IAAM,CAACR,WAAWY,GAAG,CAACJ,EAAEC,IAAI,GAAGR,GAAG,CAAC,CAACO,IAAMA,EAAEC,IAAI;IAEtF,MAAMI,YAAYd,MAAMY,MAAM,CAAC,CAACT;QAC9B,MAAMY,MAAMR,SAASS,GAAG,CAACb,EAAEZ,OAAO;QAClC,OAAO,CAACwB,OAAOA,IAAIE,MAAM,KAAKd,EAAEe,OAAO,IAAIH,IAAII,KAAK,KAAKhB,EAAEiB,IAAI;IACjE;IAEA,IAAIT,SAASU,MAAM,KAAK,KAAKP,UAAUO,MAAM,KAAK,GAAG,OAAO;QAAEC,QAAQ;QAAGC,UAAU,EAAE;IAAC;IAEtF,MAAMC,WAAWpD,eAAe0B;IAChC,MAAM2B,cAAc/C,WAAWmB;IAC/B,MAAM6B,aAAuB,EAAE;IAC/B,MAAMC,aAA0B,EAAE;IAClC,MAAMJ,WAAqB,EAAE;IAE7B,oFAAoF;IACpF,uDAAuD;IACvD,MAAMK,SAASvD,SAAS,mBAAmByC,UAAUO,MAAM;IAC3D,IAAIQ,cAAc;IAClB,KAAK,MAAMC,QAAQhB,UAAW;QAC5B,sFAAsF;QACtF,mEAAmE;QACnE,MAAMiB,eAAeP,SAASZ,MAAM,CAAC,CAACoB,UAAY,CAACA,QAAQC,cAAc,IAAID,QAAQC,cAAc,CAACnC,KAAKgC;QACzG,MAAM,EAAE1C,GAAG,EAAEmC,UAAUW,YAAY,EAAE,GAAG3D,UAAUuD,MAAMC;QACxDH,OAAOO,IAAI,CAAC,EAAEN;QACdN,SAASa,IAAI,IAAIF;QACjB,KAAK,MAAMG,OAAOC,OAAOC,IAAI,CAACnD,IAAIoD,IAAI,EAAG;YACvC,IAAI,CAACf,YAAYZ,GAAG,CAACwB,MAAM;gBACzBZ,YAAYgB,GAAG,CAACJ;gBAChBX,WAAWU,IAAI,CAACC;YAClB;QACF;QACAV,WAAWS,IAAI,CAAChD;IAClB;IACAwC,OAAOc,MAAM;IAEb,MAAMC,aAAa;WAAIlB;KAAY;IACnC,uEAAuE;IACvE,sGAAsG;IACtG,IAAIkB,WAAWtB,MAAM,GAAGrC,yBAAyB;QAC/C,MAAM,IAAIb,WACR,gBACA,CAAC,uBAAuB,EAAEwE,WAAWtB,MAAM,CAAC,0EAA0E,EAAErC,wBAAwB,wKAAwK,CAAC;IAE7T;IACA,0FAA0F;IAC1F,sEAAsE;IACtE,MAAM4D,kBAAkBD,WAAW/B,MAAM,CAAC,CAACiC,IAAM/D,yBAAyB+B,GAAG,CAACgC,MAAM,CAACrE,iBAAiBqC,GAAG,CAACgC;IAC1G,sFAAsF;IACtF,gDAAgD;IAChD,MAAMC,YAAY,CAAC,yBAAyB,EAAEF,gBAAgB1C,GAAG,CAACtB,YAAYmE,IAAI,CAAC,MAAM,UAAU,EAAEH,gBAAgB1C,GAAG,CAAC,IAAM,KAAK6C,IAAI,CAAC,MAAM,oCAAoC,EAAEH,gBAClLhC,MAAM,CAAC,CAACiC,IAAMA,MAAM,QACpB3C,GAAG,CAAC,CAAC2C,IAAM,GAAGjE,WAAWiE,GAAG,YAAY,EAAEjE,WAAWiE,IAAI,EACzDE,IAAI,CAAC,OAAO;IAEf,MAAMC,QAAQlC,UAAUF,MAAM,CAAC,CAACT,IAAM,CAACI,SAASM,GAAG,CAACV,EAAEZ,OAAO,GAAGW,GAAG,CAAC,CAACC,IAAMA,EAAEZ,OAAO;IACpF,MAAM0D,QAAwB;QAAEjD;QAAOkD,UAAUvB,WAAWzB,GAAG,CAAC,CAACiD,IAAMA,EAAE5D,OAAO;QAAGyD;QAAOrC;IAAS;IAEnG,MAAMyC,UAAUC,KAAKC,GAAG;IACxBzD,GAAG0D,IAAI,CAAC;IACR,IAAI;YAkD8BvB;QAjDhC,KAAK,MAAMwB,OAAO9B,WAAY7B,GAAG0D,IAAI,CAAC,CAAC,mCAAmC,EAAE3E,WAAW4E,MAAM;QAC7F,6EAA6E;QAC7E,kFAAkF;QAClF,+EAA+E;QAC/E,MAAMC,UAAU5D,GAAGQ,OAAO,CAAC,CAAC,kFAAkF,CAAC;QAC/G,MAAMqD,iBAAiB7D,GAAGQ,OAAO,CAAC,CAAC,yCAAyC,CAAC;QAC7E,MAAMsD,mBAAmB9D,GAAGQ,OAAO,CAAC,CAAC,uDAAuD,CAAC;QAC7F,IAAIM,SAASU,MAAM,GAAG,GAAG;YACvB,MAAMuC,MAAM/D,GAAGQ,OAAO,CAAC,CAAC,wCAAwC,CAAC;YACjE,KAAK,MAAMK,QAAQC,SAAU;oBAMKqB;gBALhC,kFAAkF;gBAClF,qEAAqE;gBACrEyB,QAAQnE,GAAG,CAACoB;gBACZkD,IAAItE,GAAG,CAACoB;gBACRgD,eAAepE,GAAG,CAACoB;gBACnB,KAAK,MAAMsB,WAAWR,UAAUQ,kBAAAA,QAAQ6B,MAAM,cAAd7B,sCAAAA,qBAAAA,SAAiBnC,IAAIa,MAAMuC;YAC7D;QACF;QACA,IAAItB,WAAWN,MAAM,GAAG,GAAG;YACzB,MAAMyC,SAASjE,GAAGQ,OAAO,CAACyC;YAC1B,MAAM3D,aAAaU,GAAGQ,OAAO,CAACpB;YAC9B,wFAAwF;YACxF,+EAA+E;YAC/E,MAAMI,aAAanB,gBAAgB4B,SAASiE;YAC5C,KAAK,MAAM3E,OAAOuC,WAAY;oBAsBIK;gBArBhC,MAAMgC,SAASpB,gBAAgB1C,GAAG,CAAC,CAACsD;wBAO3BpE;oBANP,IAAIoE,QAAQ,QAAQ,OAAOpE,IAAIG,OAAO;oBACtC,IAAIiE,QAAQ,UAAU,OAAOpE,IAAI8B,OAAO;oBACxC,IAAIsC,QAAQ,UAAU,OAAOpE,IAAI6E,OAAO;oBACxC,IAAIT,QAAQ,SAAS,OAAOpE,IAAIgC,IAAI;oBACpC,mFAAmF;oBACnF,IAAIoC,QAAQ,gBAAgB,OAAOpE,IAAI8E,UAAU;oBACjD,QAAO9E,gBAAAA,IAAIoD,IAAI,CAACgB,IAAI,cAAbpE,2BAAAA,gBAAiB;gBAC1B;gBACA,uFAAuF;gBACvF0E,OAAOxE,GAAG,IAAI0E;gBACd,oFAAoF;gBACpF,IAAIzD,SAASM,GAAG,CAACzB,IAAIG,OAAO,GAAG;wBAEGyC;oBADhCyB,QAAQnE,GAAG,CAACF,IAAIG,OAAO;oBACvB,KAAK,MAAMyC,WAAWR,UAAUQ,mBAAAA,QAAQ6B,MAAM,cAAd7B,uCAAAA,sBAAAA,SAAiBnC,IAAIT,IAAIG,OAAO,EAAE0D;gBACpE;gBACA/D,iBAAiBC,YAAYC,KAAKC;gBAClC,iFAAiF;gBACjF,2EAA2E;gBAC3E,IAAIkB,SAASM,GAAG,CAACzB,IAAIG,OAAO,GAAGmE,eAAepE,GAAG,CAACF,IAAIG,OAAO;gBAC7D,KAAK,MAAM4E,cAAc/E,IAAIgF,OAAO,CAAET,iBAAiBrE,GAAG,CAACF,IAAIG,OAAO,EAAE4E;gBACxE,KAAK,MAAMnC,WAAWR,UAAUQ,iBAAAA,QAAQqC,KAAK,cAAbrC,qCAAAA,oBAAAA,SAAgBnC,IAAIT,IAAIG,OAAO,EAAEH,IAAIkF,SAAS,CAACtC,QAAQuC,IAAI,CAAC,EAAEtB;YAChG;QACF;QACA,KAAK,MAAMjB,WAAWR,UAAUQ,0BAAAA,QAAQwC,cAAc,cAAtBxC,8CAAAA,6BAAAA,SAAyBnC,IAAIoD;QAC7DpD,GAAG0D,IAAI,CAAC;IACV,EAAE,OAAOkB,KAAK;QACZ5E,GAAG0D,IAAI,CAAC;QACR,MAAMkB;IACR;IAEA,qFAAqF;IACrF,8GAA8G;IAC9G,MAAMC,aAAarB,KAAKC,GAAG,KAAKF;IAChC,MAAMuB,UAAUhG,QAAQkB,IAAI;IAC5B,yFAAyF;IACzF,+EAA+E;IAC/E,MAAM+E,UAAUD,YAAY,OAAO,CAAC,IAAIE,OAAOF;IAC/C,IAAID,aAAaE,SAAS/F,QAAQgB,IAAI,oBAAoBiF,OAAOJ;IAEjE,OAAO;QAAEpD,QAAQK,WAAWN,MAAM;QAAEE;IAAS;AAC/C;AAEA,wFAAwF;AACxF,mFAAmF;AACnF,OAAO,SAASwD,qBAAqBC,MAAc,EAAEC,KAAa;IAChE,MAAMC,QAAQ,CAACC,OAAkBA,KAAKC,UAAU,CAAC,aAAaD,KAAKE,KAAK,CAAC,KAAKC,KAAK,CAAC,GAAG,GAAGvC,IAAI,CAAC,OAAOoC,KAAKE,KAAK,CAAC,IAAI,CAAC,EAAE;IACxH,MAAME,QAAQ,CAACC,MAAgB,IAAIhF,IAAIgF,IAAIH,KAAK,CAAC,KAAKnF,GAAG,CAAC,CAACiF,OAAS;gBAACD,MAAMC;gBAAOA;aAAK;IACvF,MAAMM,IAAIF,MAAMP;IAChB,MAAMU,IAAIH,MAAMN;IAChB,MAAMU,UAAU,IAAI5G;IACpB,KAAK,MAAM,CAACsD,KAAKuD,IAAI,IAAIF,EAAG,IAAID,EAAEzE,GAAG,CAACqB,SAASuD,KAAKD,QAAQlD,GAAG,CAACJ;IAChE,KAAK,MAAMA,OAAOoD,EAAElD,IAAI,GAAI,IAAI,CAACmD,EAAE7E,GAAG,CAACwB,MAAMsD,QAAQlD,GAAG,CAACJ;IACzD,OAAOsD;AACT;AAEA,4CAA4C;AAC5C,OAAO,SAASE,cAAcb,MAAc,EAAEC,KAAa;IACzD,MAAMU,UAAUZ,qBAAqBC,QAAQC;IAC7C,MAAMa,QAAQ,CAACzD,MAAiBA,QAAQ,UAAU,mBAAmBA,QAAQ,aAAa,sBAAsBA,IAAI+C,UAAU,CAAC,aAAa,CAAC,QAAQ,EAAE/C,IAAIiD,KAAK,CAAC,GAAG,CAAC,CAAC,GAAG;IACzK,OAAOK,QAAQvE,IAAI,KAAK,IAAI,aAAa;WAAIuE;KAAQ,CAACzF,GAAG,CAAC4F,OAAO/C,IAAI,CAAC;AACxE;AAEA,6FAA6F;AAC7F,sFAAsF;AACtF,gGAAgG;AAChG,uFAAuF;AACvF,2FAA2F;AAC3F,kGAAkG;AAClG,OAAO,SAASgD,oBAAoBlG,EAAgB,EAAEC,GAAW,EAAEC,OAAe;IAChF,MAAMiG,QAAQ,IAAIjH,IAAI,AAACc,GAAGQ,OAAO,CAAC,kCAAkCC,GAAG,GAA+BJ,GAAG,CAAC,CAACO,IAAMA,EAAEC,IAAI;IACvH,MAAMV,QAAQ1B,UAAUwB,KAAKC,SAASa,MAAM,CAAC,CAACT,IAAM6F,MAAMnF,GAAG,CAACV,EAAEZ,OAAO;IACvE,MAAMF,aAAanB,gBAAgB4B,SAASiE;IAC5C,MAAM5E,aAAaU,GAAGQ,OAAO,CAACpB;IAC9B,MAAMsC,WAAqB,EAAE;IAC7B1B,GAAG0D,IAAI,CAAC;IACR,IAAI;QACF,KAAK,MAAMzB,QAAQ9B,MAAO;YACxB,MAAM,EAAEZ,GAAG,EAAEmC,UAAUW,YAAY,EAAE,GAAG3D,UAAUuD;YAClDP,SAASa,IAAI,IAAIF;YACjBhD,iBAAiBC,YAAYC,KAAKC;QACpC;QACAQ,GAAG0D,IAAI,CAAC;IACV,EAAE,OAAOkB,KAAK;QACZ5E,GAAG0D,IAAI,CAAC;QACR,MAAMkB;IACR;IACA,OAAOlD;AACT"}
+{"version":3,"sources":["/Users/kevin/Dev/OpenSource/ai/sensemaking/src/db/reconcile.ts"],"sourcesContent":["import type { DatabaseSync, StatementSync } from 'node:sqlite';\nimport type { Config } from '../config/index.ts';\nimport { contentTokenize } from '../config/index.ts';\nimport { SenseError } from '../errors.ts';\nimport { activeFeatures } from '../features/index.ts';\nimport type { ReconcileDelta } from '../features/types.ts';\nimport { progress } from '../progress.ts';\nimport type { ParsedDoc } from '../scan.ts';\nimport { listFiles, parseFile, RESERVED_COLUMNS } from '../scan.ts';\nimport { segmentField } from '../segment.ts';\nimport { getColumns, getMeta, quoteIdent, setMeta } from './shared.ts';\n\n// Feature-owned columns (`_rank`) must stay out of the upsert: a reparse would null the last\n// computed value on every touch, not just the reconciles that recompute it.\nconst CORE_FRONTMATTER_COLUMNS = new Set(['path', '_mtime', '_ctime', '_size', '_parse_error']);\n\n// SQLite's compile-time SQLITE_MAX_COLUMN, default 2000 (https://www.sqlite.org/limits.html).\nconst MAX_FRONTMATTER_COLUMNS = 2000;\n\n// Shared by reconcile() and the tokenize-only rebuild in open(), so both prepare the same\n// literal instead of two copies drifting apart.\nconst INSERT_CONTENT_SQL = `INSERT INTO content (rowid, title, summary, text, \"path\", title_seg, summary_seg, text_seg) VALUES ((SELECT rowid FROM frontmatter WHERE \"path\" = ?), ?, ?, ?, ?, ?, ?, ?)`;\n\n// Shared by reconcile() and the tokenize-only rebuild in open(), so content population is\n// defined once. Assumes the frontmatter row for doc.relPath already exists (rowid lookup).\nfunction insertContentRow(insertBody: StatementSync, doc: ParsedDoc, segmenting: boolean): void {\n  insertBody.run(doc.relPath, doc.search.title, doc.search.summary, doc.search.text, doc.relPath, segmenting ? segmentField(doc.search.title) : '', segmenting ? segmentField(doc.search.summary) : '', segmenting ? segmentField(doc.search.text) : '');\n}\n\nexport function reconcile(db: DatabaseSync, cfg: Config, baseDir: string): { parsed: number; warnings: string[] } {\n  const files = listFiles(cfg, baseDir);\n  const currentSet = new Set(files.map((f) => f.relPath));\n\n  const existingRows = db.prepare(`SELECT \"path\", \"_mtime\", \"_size\" FROM frontmatter`).all() as Array<{\n    path: string;\n    _mtime: number;\n    _size: number;\n  }>;\n  const existing = new Map(existingRows.map((r) => [r.path, r]));\n  const vanished = existingRows.filter((r) => !currentSet.has(r.path)).map((r) => r.path);\n\n  const toReparse = files.filter((f) => {\n    const row = existing.get(f.relPath);\n    return !row || row._mtime !== f.mtimeMs || row._size !== f.size;\n  });\n\n  if (vanished.length === 0 && toReparse.length === 0) return { parsed: 0, warnings: [] };\n\n  const features = activeFeatures(cfg);\n  const seenColumns = getColumns(db);\n  const newColumns: string[] = [];\n  const parsedDocs: ParsedDoc[] = [];\n  const warnings: string[] = [];\n\n  // Bulk reparses (a sync, a cold build) are the long silences a query can hit; short\n  // reconciles stay silent (progress() has a threshold).\n  const report = progress('reparsing files', toReparse.length);\n  let parsedCount = 0;\n  for (const file of toReparse) {\n    // A doc only gets extract/store from features that apply to it (currently: embed, via\n    // FileStat.embed -- true iff the config names an embedding model).\n    const fileFeatures = features.filter((feature) => !feature.enabledForFile || feature.enabledForFile(cfg, file));\n    const { doc, warnings: fileWarnings } = parseFile(file, fileFeatures);\n    report.tick(++parsedCount);\n    warnings.push(...fileWarnings);\n    for (const key of Object.keys(doc.data)) {\n      if (!seenColumns.has(key)) {\n        seenColumns.add(key);\n        newColumns.push(key);\n      }\n    }\n    parsedDocs.push(doc);\n  }\n  report.finish();\n\n  const allColumns = [...seenColumns];\n  // Fence before ALTERing: SQLite's own failure past this point is a raw\n  // \"too many columns on sqlite_altertab_frontmatter\" with no indication of the boundary or the levers.\n  if (allColumns.length > MAX_FRONTMATTER_COLUMNS) {\n    throw new SenseError(\n      'COLUMN_LIMIT',\n      `frontmatter would need ${allColumns.length} columns, crossing SQLite's compile-time SQLITE_MAX_COLUMN limit (default ${MAX_FRONTMATTER_COLUMNS}; see https://www.sqlite.org/limits.html). Narrow the presets' include globs so fewer/other files are indexed, or fix whatever is generating unbounded frontmatter keys.`\n    );\n  }\n  // Columns the frontmatter upsert actually writes: core + parsed frontmatter keys, never a\n  // feature-owned reserved column (see CORE_FRONTMATTER_COLUMNS above).\n  const writableColumns = allColumns.filter((c) => CORE_FRONTMATTER_COLUMNS.has(c) || !RESERVED_COLUMNS.has(c));\n  // ON CONFLICT UPDATE (not OR REPLACE) keeps the row's rowid stable across reparses --\n  // content rows are coupled to that rowid below.\n  const insertSql = `INSERT INTO frontmatter (${writableColumns.map(quoteIdent).join(', ')}) VALUES (${writableColumns.map(() => '?').join(', ')}) ON CONFLICT(\"path\") DO UPDATE SET ${writableColumns\n    .filter((c) => c !== 'path')\n    .map((c) => `${quoteIdent(c)} = excluded.${quoteIdent(c)}`)\n    .join(', ')}`;\n\n  const added = toReparse.filter((f) => !existing.has(f.relPath)).map((f) => f.relPath);\n  const delta: ReconcileDelta = { files, reparsed: parsedDocs.map((d) => d.relPath), added, vanished };\n\n  const txStart = Date.now();\n  db.exec('BEGIN');\n  try {\n    for (const col of newColumns) db.exec(`ALTER TABLE frontmatter ADD COLUMN ${quoteIdent(col)}`);\n    // FTS5 has no upsert, so delete-before-insert into `content`; coupled to the\n    // frontmatter rowid (indexed via its PRIMARY KEY) instead of the UNINDEXED `path`\n    // column, which a per-row DELETE would otherwise scan the whole table to find.\n    const delBody = db.prepare(`DELETE FROM content WHERE rowid = (SELECT rowid FROM frontmatter WHERE \"path\" = ?)`);\n    const delPresetFiles = db.prepare(`DELETE FROM preset_files WHERE \"path\" = ?`);\n    const insertPresetFile = db.prepare(`INSERT INTO preset_files (\"path\", preset) VALUES (?, ?)`);\n    if (vanished.length > 0) {\n      const del = db.prepare(`DELETE FROM frontmatter WHERE \"path\" = ?`);\n      for (const path of vanished) {\n        // content delete must run first: it looks up the frontmatter rowid by path, which\n        // the frontmatter delete below would otherwise have already removed.\n        delBody.run(path);\n        del.run(path);\n        delPresetFiles.run(path);\n        for (const feature of features) feature.remove?.(db, path, delta);\n      }\n    }\n    if (parsedDocs.length > 0) {\n      const insert = db.prepare(insertSql);\n      const insertBody = db.prepare(INSERT_CONTENT_SQL);\n      // A non-default tokenizer means the tree has chosen its own scheme; a phrase query over\n      // grapheme runs would be nonsense against trigram, so the sidecars stay empty.\n      const segmenting = contentTokenize(cfg) === undefined;\n      for (const doc of parsedDocs) {\n        const values = writableColumns.map((col) => {\n          if (col === 'path') return doc.relPath;\n          if (col === '_mtime') return doc.mtimeMs;\n          if (col === '_ctime') return doc.ctimeMs;\n          if (col === '_size') return doc.size;\n          // Written per parse, unlike _rank, which a feature pass owns and the upsert skips.\n          if (col === '_parse_error') return doc.parseError;\n          return doc.data[col] ?? null;\n        });\n        // Frontmatter upsert first: content's rowid lookup below depends on this row existing.\n        insert.run(...values);\n        // Only for docs that have rows: an unconditional delete made cold crawls quadratic.\n        if (existing.has(doc.relPath)) {\n          delBody.run(doc.relPath);\n          for (const feature of features) feature.remove?.(db, doc.relPath, delta);\n        }\n        insertContentRow(insertBody, doc, segmenting);\n        // A preset edit forces a full rebuild, so an unchanged doc's coverage is already\n        // correct; new docs have nothing to clear, which keeps cold builds linear.\n        if (existing.has(doc.relPath)) delPresetFiles.run(doc.relPath);\n        for (const presetName of doc.presets) insertPresetFile.run(doc.relPath, presetName);\n        for (const feature of features) feature.store?.(db, doc.relPath, doc.extracted[feature.name], delta);\n      }\n    }\n    for (const feature of features) feature.afterReconcile?.(db, delta);\n    db.exec('COMMIT');\n  } catch (err) {\n    db.exec('ROLLBACK');\n    throw err;\n  }\n\n  // Reconcile's own write-transaction duration, for open()'s derived busy_timeout (F):\n  // keep the observed max so a big watcher reconcile's lock hold is what the next open bounds its wait against.\n  const durationMs = Date.now() - txStart;\n  const prevRaw = getMeta(db, 'reconcile_max_ms');\n  // -1, not 0, so a genuinely 0ms first reconcile (sub-millisecond, common on a tiny tree)\n  // still gets recorded instead of losing to the \"nothing recorded yet\" default.\n  const prevMax = prevRaw === null ? -1 : Number(prevRaw);\n  if (durationMs > prevMax) setMeta(db, 'reconcile_max_ms', String(durationMs));\n\n  return { parsed: parsedDocs.length, warnings };\n}\n\n// Segment keys that moved between two feature signatures (see config.featureSignature's\n// format: global features, embed provider, tokenize, then one segment per preset).\nexport function changedSignatureKeys(before: string, after: string): Set<string> {\n  const keyOf = (part: string) => (part.startsWith('preset:') ? part.split(':').slice(0, 2).join(':') : part.split(':')[0]);\n  const parse = (sig: string) => new Map(sig.split('|').map((part) => [keyOf(part), part]));\n  const a = parse(before);\n  const b = parse(after);\n  const changed = new Set<string>();\n  for (const [key, val] of b) if (a.get(key) !== val) changed.add(key);\n  for (const key of a.keys()) if (!b.has(key)) changed.add(key);\n  return changed;\n}\n\n// Whether the embed segment only gained its resolved weight identity: same provider and\n// model, no identity recorded before, one now -- adopted into meta without a rebuild.\nexport function embedIdentityAdopted(before: string, after: string): boolean {\n  const embedPart = (sig: string) => sig.split('|').find((p) => p.startsWith('embed:'));\n  const b = embedPart(before);\n  const a = embedPart(after);\n  if (b === undefined || a === undefined) return false;\n  const at = a.indexOf('@');\n  return b.indexOf('@') === -1 && at !== -1 && a.slice(0, at) === b;\n}\n\n// Names what moved, for the rebuild notice.\nexport function signatureDiff(before: string, after: string): string {\n  const changed = changedSignatureKeys(before, after);\n  const label = (key: string) => (key === 'embed' ? 'embed settings' : key === 'tokenize' ? 'content tokenizer' : key.startsWith('preset:') ? `preset \"${key.slice(7)}\"` : 'features');\n  return changed.size === 0 ? 'features' : [...changed].map(label).join(', ');\n}\n\n// The tokenize-only rebuild in open(): content is dropped and repopulated from files already\n// listed in frontmatter, which itself is untouched. Frontmatter, links, sections, and\n// embeddings are file-derived and tokenizer-independent, so they survive. No feature extractors\n// run here -- doc.search (title/summary/text) is all content population needs. Returns\n// parseFile's per-file warnings (e.g. a bad date) so open() can surface them -- mtimes are\n// untouched, so reconcile() never reparses these files and would otherwise never emit them again.\nexport function rebuildContentTable(db: DatabaseSync, cfg: Config, baseDir: string): string[] {\n  const known = new Set((db.prepare('SELECT \"path\" FROM frontmatter').all() as Array<{ path: string }>).map((r) => r.path));\n  const files = listFiles(cfg, baseDir).filter((f) => known.has(f.relPath));\n  const segmenting = contentTokenize(cfg) === undefined;\n  const insertBody = db.prepare(INSERT_CONTENT_SQL);\n  const warnings: string[] = [];\n  db.exec('BEGIN');\n  try {\n    for (const file of files) {\n      const { doc, warnings: fileWarnings } = parseFile(file);\n      warnings.push(...fileWarnings);\n      insertContentRow(insertBody, doc, segmenting);\n    }\n    db.exec('COMMIT');\n  } catch (err) {\n    db.exec('ROLLBACK');\n    throw err;\n  }\n  return warnings;\n}\n"],"names":["contentTokenize","SenseError","activeFeatures","progress","listFiles","parseFile","RESERVED_COLUMNS","segmentField","getColumns","getMeta","quoteIdent","setMeta","CORE_FRONTMATTER_COLUMNS","Set","MAX_FRONTMATTER_COLUMNS","INSERT_CONTENT_SQL","insertContentRow","insertBody","doc","segmenting","run","relPath","search","title","summary","text","reconcile","db","cfg","baseDir","files","currentSet","map","f","existingRows","prepare","all","existing","Map","r","path","vanished","filter","has","toReparse","row","get","_mtime","mtimeMs","_size","size","length","parsed","warnings","features","seenColumns","newColumns","parsedDocs","report","parsedCount","file","fileFeatures","feature","enabledForFile","fileWarnings","tick","push","key","Object","keys","data","add","finish","allColumns","writableColumns","c","insertSql","join","added","delta","reparsed","d","txStart","Date","now","exec","col","delBody","delPresetFiles","insertPresetFile","del","remove","insert","undefined","values","ctimeMs","parseError","presetName","presets","store","extracted","name","afterReconcile","err","durationMs","prevRaw","prevMax","Number","String","changedSignatureKeys","before","after","keyOf","part","startsWith","split","slice","parse","sig","a","b","changed","val","embedIdentityAdopted","embedPart","find","p","at","indexOf","signatureDiff","label","rebuildContentTable","known"],"mappings":"AAEA,SAASA,eAAe,QAAQ,qBAAqB;AACrD,SAASC,UAAU,QAAQ,eAAe;AAC1C,SAASC,cAAc,QAAQ,uBAAuB;AAEtD,SAASC,QAAQ,QAAQ,iBAAiB;AAE1C,SAASC,SAAS,EAAEC,SAAS,EAAEC,gBAAgB,QAAQ,aAAa;AACpE,SAASC,YAAY,QAAQ,gBAAgB;AAC7C,SAASC,UAAU,EAAEC,OAAO,EAAEC,UAAU,EAAEC,OAAO,QAAQ,cAAc;AAEvE,6FAA6F;AAC7F,4EAA4E;AAC5E,MAAMC,2BAA2B,IAAIC,IAAI;IAAC;IAAQ;IAAU;IAAU;IAAS;CAAe;AAE9F,8FAA8F;AAC9F,MAAMC,0BAA0B;AAEhC,0FAA0F;AAC1F,gDAAgD;AAChD,MAAMC,qBAAqB,CAAC,0KAA0K,CAAC;AAEvM,0FAA0F;AAC1F,2FAA2F;AAC3F,SAASC,iBAAiBC,UAAyB,EAAEC,GAAc,EAAEC,UAAmB;IACtFF,WAAWG,GAAG,CAACF,IAAIG,OAAO,EAAEH,IAAII,MAAM,CAACC,KAAK,EAAEL,IAAII,MAAM,CAACE,OAAO,EAAEN,IAAII,MAAM,CAACG,IAAI,EAAEP,IAAIG,OAAO,EAAEF,aAAaZ,aAAaW,IAAII,MAAM,CAACC,KAAK,IAAI,IAAIJ,aAAaZ,aAAaW,IAAII,MAAM,CAACE,OAAO,IAAI,IAAIL,aAAaZ,aAAaW,IAAII,MAAM,CAACG,IAAI,IAAI;AACrP;AAEA,OAAO,SAASC,UAAUC,EAAgB,EAAEC,GAAW,EAAEC,OAAe;IACtE,MAAMC,QAAQ1B,UAAUwB,KAAKC;IAC7B,MAAME,aAAa,IAAIlB,IAAIiB,MAAME,GAAG,CAAC,CAACC,IAAMA,EAAEZ,OAAO;IAErD,MAAMa,eAAeP,GAAGQ,OAAO,CAAC,CAAC,iDAAiD,CAAC,EAAEC,GAAG;IAKxF,MAAMC,WAAW,IAAIC,IAAIJ,aAAaF,GAAG,CAAC,CAACO,IAAM;YAACA,EAAEC,IAAI;YAAED;SAAE;IAC5D,MAAME,WAAWP,aAAaQ,MAAM,CAAC,CAACH,IAAM,CAACR,WAAWY,GAAG,CAACJ,EAAEC,IAAI,GAAGR,GAAG,CAAC,CAACO,IAAMA,EAAEC,IAAI;IAEtF,MAAMI,YAAYd,MAAMY,MAAM,CAAC,CAACT;QAC9B,MAAMY,MAAMR,SAASS,GAAG,CAACb,EAAEZ,OAAO;QAClC,OAAO,CAACwB,OAAOA,IAAIE,MAAM,KAAKd,EAAEe,OAAO,IAAIH,IAAII,KAAK,KAAKhB,EAAEiB,IAAI;IACjE;IAEA,IAAIT,SAASU,MAAM,KAAK,KAAKP,UAAUO,MAAM,KAAK,GAAG,OAAO;QAAEC,QAAQ;QAAGC,UAAU,EAAE;IAAC;IAEtF,MAAMC,WAAWpD,eAAe0B;IAChC,MAAM2B,cAAc/C,WAAWmB;IAC/B,MAAM6B,aAAuB,EAAE;IAC/B,MAAMC,aAA0B,EAAE;IAClC,MAAMJ,WAAqB,EAAE;IAE7B,oFAAoF;IACpF,uDAAuD;IACvD,MAAMK,SAASvD,SAAS,mBAAmByC,UAAUO,MAAM;IAC3D,IAAIQ,cAAc;IAClB,KAAK,MAAMC,QAAQhB,UAAW;QAC5B,sFAAsF;QACtF,mEAAmE;QACnE,MAAMiB,eAAeP,SAASZ,MAAM,CAAC,CAACoB,UAAY,CAACA,QAAQC,cAAc,IAAID,QAAQC,cAAc,CAACnC,KAAKgC;QACzG,MAAM,EAAE1C,GAAG,EAAEmC,UAAUW,YAAY,EAAE,GAAG3D,UAAUuD,MAAMC;QACxDH,OAAOO,IAAI,CAAC,EAAEN;QACdN,SAASa,IAAI,IAAIF;QACjB,KAAK,MAAMG,OAAOC,OAAOC,IAAI,CAACnD,IAAIoD,IAAI,EAAG;YACvC,IAAI,CAACf,YAAYZ,GAAG,CAACwB,MAAM;gBACzBZ,YAAYgB,GAAG,CAACJ;gBAChBX,WAAWU,IAAI,CAACC;YAClB;QACF;QACAV,WAAWS,IAAI,CAAChD;IAClB;IACAwC,OAAOc,MAAM;IAEb,MAAMC,aAAa;WAAIlB;KAAY;IACnC,uEAAuE;IACvE,sGAAsG;IACtG,IAAIkB,WAAWtB,MAAM,GAAGrC,yBAAyB;QAC/C,MAAM,IAAIb,WACR,gBACA,CAAC,uBAAuB,EAAEwE,WAAWtB,MAAM,CAAC,0EAA0E,EAAErC,wBAAwB,wKAAwK,CAAC;IAE7T;IACA,0FAA0F;IAC1F,sEAAsE;IACtE,MAAM4D,kBAAkBD,WAAW/B,MAAM,CAAC,CAACiC,IAAM/D,yBAAyB+B,GAAG,CAACgC,MAAM,CAACrE,iBAAiBqC,GAAG,CAACgC;IAC1G,sFAAsF;IACtF,gDAAgD;IAChD,MAAMC,YAAY,CAAC,yBAAyB,EAAEF,gBAAgB1C,GAAG,CAACtB,YAAYmE,IAAI,CAAC,MAAM,UAAU,EAAEH,gBAAgB1C,GAAG,CAAC,IAAM,KAAK6C,IAAI,CAAC,MAAM,oCAAoC,EAAEH,gBAClLhC,MAAM,CAAC,CAACiC,IAAMA,MAAM,QACpB3C,GAAG,CAAC,CAAC2C,IAAM,GAAGjE,WAAWiE,GAAG,YAAY,EAAEjE,WAAWiE,IAAI,EACzDE,IAAI,CAAC,OAAO;IAEf,MAAMC,QAAQlC,UAAUF,MAAM,CAAC,CAACT,IAAM,CAACI,SAASM,GAAG,CAACV,EAAEZ,OAAO,GAAGW,GAAG,CAAC,CAACC,IAAMA,EAAEZ,OAAO;IACpF,MAAM0D,QAAwB;QAAEjD;QAAOkD,UAAUvB,WAAWzB,GAAG,CAAC,CAACiD,IAAMA,EAAE5D,OAAO;QAAGyD;QAAOrC;IAAS;IAEnG,MAAMyC,UAAUC,KAAKC,GAAG;IACxBzD,GAAG0D,IAAI,CAAC;IACR,IAAI;YAkD8BvB;QAjDhC,KAAK,MAAMwB,OAAO9B,WAAY7B,GAAG0D,IAAI,CAAC,CAAC,mCAAmC,EAAE3E,WAAW4E,MAAM;QAC7F,6EAA6E;QAC7E,kFAAkF;QAClF,+EAA+E;QAC/E,MAAMC,UAAU5D,GAAGQ,OAAO,CAAC,CAAC,kFAAkF,CAAC;QAC/G,MAAMqD,iBAAiB7D,GAAGQ,OAAO,CAAC,CAAC,yCAAyC,CAAC;QAC7E,MAAMsD,mBAAmB9D,GAAGQ,OAAO,CAAC,CAAC,uDAAuD,CAAC;QAC7F,IAAIM,SAASU,MAAM,GAAG,GAAG;YACvB,MAAMuC,MAAM/D,GAAGQ,OAAO,CAAC,CAAC,wCAAwC,CAAC;YACjE,KAAK,MAAMK,QAAQC,SAAU;oBAMKqB;gBALhC,kFAAkF;gBAClF,qEAAqE;gBACrEyB,QAAQnE,GAAG,CAACoB;gBACZkD,IAAItE,GAAG,CAACoB;gBACRgD,eAAepE,GAAG,CAACoB;gBACnB,KAAK,MAAMsB,WAAWR,UAAUQ,kBAAAA,QAAQ6B,MAAM,cAAd7B,sCAAAA,qBAAAA,SAAiBnC,IAAIa,MAAMuC;YAC7D;QACF;QACA,IAAItB,WAAWN,MAAM,GAAG,GAAG;YACzB,MAAMyC,SAASjE,GAAGQ,OAAO,CAACyC;YAC1B,MAAM3D,aAAaU,GAAGQ,OAAO,CAACpB;YAC9B,wFAAwF;YACxF,+EAA+E;YAC/E,MAAMI,aAAanB,gBAAgB4B,SAASiE;YAC5C,KAAK,MAAM3E,OAAOuC,WAAY;oBAsBIK;gBArBhC,MAAMgC,SAASpB,gBAAgB1C,GAAG,CAAC,CAACsD;wBAO3BpE;oBANP,IAAIoE,QAAQ,QAAQ,OAAOpE,IAAIG,OAAO;oBACtC,IAAIiE,QAAQ,UAAU,OAAOpE,IAAI8B,OAAO;oBACxC,IAAIsC,QAAQ,UAAU,OAAOpE,IAAI6E,OAAO;oBACxC,IAAIT,QAAQ,SAAS,OAAOpE,IAAIgC,IAAI;oBACpC,mFAAmF;oBACnF,IAAIoC,QAAQ,gBAAgB,OAAOpE,IAAI8E,UAAU;oBACjD,QAAO9E,gBAAAA,IAAIoD,IAAI,CAACgB,IAAI,cAAbpE,2BAAAA,gBAAiB;gBAC1B;gBACA,uFAAuF;gBACvF0E,OAAOxE,GAAG,IAAI0E;gBACd,oFAAoF;gBACpF,IAAIzD,SAASM,GAAG,CAACzB,IAAIG,OAAO,GAAG;wBAEGyC;oBADhCyB,QAAQnE,GAAG,CAACF,IAAIG,OAAO;oBACvB,KAAK,MAAMyC,WAAWR,UAAUQ,mBAAAA,QAAQ6B,MAAM,cAAd7B,uCAAAA,sBAAAA,SAAiBnC,IAAIT,IAAIG,OAAO,EAAE0D;gBACpE;gBACA/D,iBAAiBC,YAAYC,KAAKC;gBAClC,iFAAiF;gBACjF,2EAA2E;gBAC3E,IAAIkB,SAASM,GAAG,CAACzB,IAAIG,OAAO,GAAGmE,eAAepE,GAAG,CAACF,IAAIG,OAAO;gBAC7D,KAAK,MAAM4E,cAAc/E,IAAIgF,OAAO,CAAET,iBAAiBrE,GAAG,CAACF,IAAIG,OAAO,EAAE4E;gBACxE,KAAK,MAAMnC,WAAWR,UAAUQ,iBAAAA,QAAQqC,KAAK,cAAbrC,qCAAAA,oBAAAA,SAAgBnC,IAAIT,IAAIG,OAAO,EAAEH,IAAIkF,SAAS,CAACtC,QAAQuC,IAAI,CAAC,EAAEtB;YAChG;QACF;QACA,KAAK,MAAMjB,WAAWR,UAAUQ,0BAAAA,QAAQwC,cAAc,cAAtBxC,8CAAAA,6BAAAA,SAAyBnC,IAAIoD;QAC7DpD,GAAG0D,IAAI,CAAC;IACV,EAAE,OAAOkB,KAAK;QACZ5E,GAAG0D,IAAI,CAAC;QACR,MAAMkB;IACR;IAEA,qFAAqF;IACrF,8GAA8G;IAC9G,MAAMC,aAAarB,KAAKC,GAAG,KAAKF;IAChC,MAAMuB,UAAUhG,QAAQkB,IAAI;IAC5B,yFAAyF;IACzF,+EAA+E;IAC/E,MAAM+E,UAAUD,YAAY,OAAO,CAAC,IAAIE,OAAOF;IAC/C,IAAID,aAAaE,SAAS/F,QAAQgB,IAAI,oBAAoBiF,OAAOJ;IAEjE,OAAO;QAAEpD,QAAQK,WAAWN,MAAM;QAAEE;IAAS;AAC/C;AAEA,wFAAwF;AACxF,mFAAmF;AACnF,OAAO,SAASwD,qBAAqBC,MAAc,EAAEC,KAAa;IAChE,MAAMC,QAAQ,CAACC,OAAkBA,KAAKC,UAAU,CAAC,aAAaD,KAAKE,KAAK,CAAC,KAAKC,KAAK,CAAC,GAAG,GAAGvC,IAAI,CAAC,OAAOoC,KAAKE,KAAK,CAAC,IAAI,CAAC,EAAE;IACxH,MAAME,QAAQ,CAACC,MAAgB,IAAIhF,IAAIgF,IAAIH,KAAK,CAAC,KAAKnF,GAAG,CAAC,CAACiF,OAAS;gBAACD,MAAMC;gBAAOA;aAAK;IACvF,MAAMM,IAAIF,MAAMP;IAChB,MAAMU,IAAIH,MAAMN;IAChB,MAAMU,UAAU,IAAI5G;IACpB,KAAK,MAAM,CAACsD,KAAKuD,IAAI,IAAIF,EAAG,IAAID,EAAEzE,GAAG,CAACqB,SAASuD,KAAKD,QAAQlD,GAAG,CAACJ;IAChE,KAAK,MAAMA,OAAOoD,EAAElD,IAAI,GAAI,IAAI,CAACmD,EAAE7E,GAAG,CAACwB,MAAMsD,QAAQlD,GAAG,CAACJ;IACzD,OAAOsD;AACT;AAEA,wFAAwF;AACxF,sFAAsF;AACtF,OAAO,SAASE,qBAAqBb,MAAc,EAAEC,KAAa;IAChE,MAAMa,YAAY,CAACN,MAAgBA,IAAIH,KAAK,CAAC,KAAKU,IAAI,CAAC,CAACC,IAAMA,EAAEZ,UAAU,CAAC;IAC3E,MAAMM,IAAII,UAAUd;IACpB,MAAMS,IAAIK,UAAUb;IACpB,IAAIS,MAAM3B,aAAa0B,MAAM1B,WAAW,OAAO;IAC/C,MAAMkC,KAAKR,EAAES,OAAO,CAAC;IACrB,OAAOR,EAAEQ,OAAO,CAAC,SAAS,CAAC,KAAKD,OAAO,CAAC,KAAKR,EAAEH,KAAK,CAAC,GAAGW,QAAQP;AAClE;AAEA,4CAA4C;AAC5C,OAAO,SAASS,cAAcnB,MAAc,EAAEC,KAAa;IACzD,MAAMU,UAAUZ,qBAAqBC,QAAQC;IAC7C,MAAMmB,QAAQ,CAAC/D,MAAiBA,QAAQ,UAAU,mBAAmBA,QAAQ,aAAa,sBAAsBA,IAAI+C,UAAU,CAAC,aAAa,CAAC,QAAQ,EAAE/C,IAAIiD,KAAK,CAAC,GAAG,CAAC,CAAC,GAAG;IACzK,OAAOK,QAAQvE,IAAI,KAAK,IAAI,aAAa;WAAIuE;KAAQ,CAACzF,GAAG,CAACkG,OAAOrD,IAAI,CAAC;AACxE;AAEA,6FAA6F;AAC7F,sFAAsF;AACtF,gGAAgG;AAChG,uFAAuF;AACvF,2FAA2F;AAC3F,kGAAkG;AAClG,OAAO,SAASsD,oBAAoBxG,EAAgB,EAAEC,GAAW,EAAEC,OAAe;IAChF,MAAMuG,QAAQ,IAAIvH,IAAI,AAACc,GAAGQ,OAAO,CAAC,kCAAkCC,GAAG,GAA+BJ,GAAG,CAAC,CAACO,IAAMA,EAAEC,IAAI;IACvH,MAAMV,QAAQ1B,UAAUwB,KAAKC,SAASa,MAAM,CAAC,CAACT,IAAMmG,MAAMzF,GAAG,CAACV,EAAEZ,OAAO;IACvE,MAAMF,aAAanB,gBAAgB4B,SAASiE;IAC5C,MAAM5E,aAAaU,GAAGQ,OAAO,CAACpB;IAC9B,MAAMsC,WAAqB,EAAE;IAC7B1B,GAAG0D,IAAI,CAAC;IACR,IAAI;QACF,KAAK,MAAMzB,QAAQ9B,MAAO;YACxB,MAAM,EAAEZ,GAAG,EAAEmC,UAAUW,YAAY,EAAE,GAAG3D,UAAUuD;YAClDP,SAASa,IAAI,IAAIF;YACjBhD,iBAAiBC,YAAYC,KAAKC;QACpC;QACAQ,GAAG0D,IAAI,CAAC;IACV,EAAE,OAAOkB,KAAK;QACZ5E,GAAG0D,IAAI,CAAC;QACR,MAAMkB;IACR;IACA,OAAOlD;AACT"}

package/dist/esm/embed/cohere.d.ts

@@ -0,0 +1,2 @@
+import type { EmbedProvider } from './types.js';
+export declare function cohereProvider(model: string, url: string | undefined, keyEnv: string | undefined): Promise<EmbedProvider>;

package/dist/esm/embed/cohere.js

@@ -0,0 +1,46 @@
+import { SenseError } from '../errors.js';
+import { fetchWithRetry } from './http.js';
+const BATCH_CAP = 96;
+// Cohere's native /v2/embed: input_type distinguishes doc vs query embeddings, which the
+// OpenAI-compatible shape cannot express.
+export async function cohereProvider(model, url, keyEnv) {
+    const base = (url !== null && url !== void 0 ? url : 'https://api.cohere.com').replace(/\/+$/, '');
+    const headers = {
+        'content-type': 'application/json'
+    };
+    const key = keyEnv ? process.env[keyEnv] : undefined;
+    if (key) headers.authorization = `Bearer ${key}`;
+    async function post(texts, inputType) {
+        const res = await fetchWithRetry(`${base}/v2/embed`, {
+            method: 'POST',
+            headers,
+            body: JSON.stringify({
+                model,
+                input_type: inputType,
+                texts,
+                embedding_types: [
+                    'float'
+                ]
+            })
+        });
+        if (!res.ok) throw new SenseError('EMBED_MODEL', `${base}/v2/embed -> HTTP ${res.status}`);
+        const body = await res.json();
+        // Cohere returns one vector per text in order, with no index field to re-sort by; the
+        // count is what a caller mapping positionally still has to be able to trust.
+        const vectors = body.embeddings.float;
+        if (vectors.length !== texts.length) throw new SenseError('EMBED_MODEL', `${base}/v2/embed returned ${vectors.length} embeddings for ${texts.length} inputs`);
+        return vectors.map((v)=>Float32Array.from(v));
+    }
+    const dims = (await post([
+        'dimension probe'
+    ], 'search_query'))[0].length;
+    return {
+        id: `cohere:${base}:${model}`,
+        dims,
+        batchCap: BATCH_CAP,
+        embedDocuments: (texts)=>post(texts, 'search_document'),
+        embedQuery: async (text)=>(await post([
+                text
+            ], 'search_query'))[0]
+    };
+}

package/dist/esm/embed/cohere.js.map

@@ -0,0 +1 @@
+{"version":3,"sources":["/Users/kevin/Dev/OpenSource/ai/sensemaking/src/embed/cohere.ts"],"sourcesContent":["import { SenseError } from '../errors.ts';\nimport { fetchWithRetry } from './http.ts';\nimport type { EmbedProvider } from './types.ts';\n\nconst BATCH_CAP = 96;\n\n// Cohere's native /v2/embed: input_type distinguishes doc vs query embeddings, which the\n// OpenAI-compatible shape cannot express.\nexport async function cohereProvider(model: string, url: string | undefined, keyEnv: string | undefined): Promise<EmbedProvider> {\n  const base = (url ?? 'https://api.cohere.com').replace(/\\/+$/, '');\n  const headers: Record<string, string> = { 'content-type': 'application/json' };\n  const key = keyEnv ? process.env[keyEnv] : undefined;\n  if (key) headers.authorization = `Bearer ${key}`;\n\n  async function post(texts: string[], inputType: 'search_document' | 'search_query'): Promise<Float32Array[]> {\n    const res = await fetchWithRetry(`${base}/v2/embed`, { method: 'POST', headers, body: JSON.stringify({ model, input_type: inputType, texts, embedding_types: ['float'] }) });\n    if (!res.ok) throw new SenseError('EMBED_MODEL', `${base}/v2/embed -> HTTP ${res.status}`);\n    const body = (await res.json()) as { embeddings: { float: number[][] } };\n    // Cohere returns one vector per text in order, with no index field to re-sort by; the\n    // count is what a caller mapping positionally still has to be able to trust.\n    const vectors = body.embeddings.float;\n    if (vectors.length !== texts.length) throw new SenseError('EMBED_MODEL', `${base}/v2/embed returned ${vectors.length} embeddings for ${texts.length} inputs`);\n    return vectors.map((v) => Float32Array.from(v));\n  }\n\n  const dims = (await post(['dimension probe'], 'search_query'))[0].length;\n  return { id: `cohere:${base}:${model}`, dims, batchCap: BATCH_CAP, embedDocuments: (texts) => post(texts, 'search_document'), embedQuery: async (text) => (await post([text], 'search_query'))[0] };\n}\n"],"names":["SenseError","fetchWithRetry","BATCH_CAP","cohereProvider","model","url","keyEnv","base","replace","headers","key","process","env","undefined","authorization","post","texts","inputType","res","method","body","JSON","stringify","input_type","embedding_types","ok","status","json","vectors","embeddings","float","length","map","v","Float32Array","from","dims","id","batchCap","embedDocuments","embedQuery","text"],"mappings":"AAAA,SAASA,UAAU,QAAQ,eAAe;AAC1C,SAASC,cAAc,QAAQ,YAAY;AAG3C,MAAMC,YAAY;AAElB,yFAAyF;AACzF,0CAA0C;AAC1C,OAAO,eAAeC,eAAeC,KAAa,EAAEC,GAAuB,EAAEC,MAA0B;IACrG,MAAMC,OAAO,CAACF,gBAAAA,iBAAAA,MAAO,0BAA0BG,OAAO,CAAC,QAAQ;IAC/D,MAAMC,UAAkC;QAAE,gBAAgB;IAAmB;IAC7E,MAAMC,MAAMJ,SAASK,QAAQC,GAAG,CAACN,OAAO,GAAGO;IAC3C,IAAIH,KAAKD,QAAQK,aAAa,GAAG,CAAC,OAAO,EAAEJ,KAAK;IAEhD,eAAeK,KAAKC,KAAe,EAAEC,SAA6C;QAChF,MAAMC,MAAM,MAAMjB,eAAe,GAAGM,KAAK,SAAS,CAAC,EAAE;YAAEY,QAAQ;YAAQV;YAASW,MAAMC,KAAKC,SAAS,CAAC;gBAAElB;gBAAOmB,YAAYN;gBAAWD;gBAAOQ,iBAAiB;oBAAC;iBAAQ;YAAC;QAAG;QAC1K,IAAI,CAACN,IAAIO,EAAE,EAAE,MAAM,IAAIzB,WAAW,eAAe,GAAGO,KAAK,kBAAkB,EAAEW,IAAIQ,MAAM,EAAE;QACzF,MAAMN,OAAQ,MAAMF,IAAIS,IAAI;QAC5B,sFAAsF;QACtF,6EAA6E;QAC7E,MAAMC,UAAUR,KAAKS,UAAU,CAACC,KAAK;QACrC,IAAIF,QAAQG,MAAM,KAAKf,MAAMe,MAAM,EAAE,MAAM,IAAI/B,WAAW,eAAe,GAAGO,KAAK,mBAAmB,EAAEqB,QAAQG,MAAM,CAAC,gBAAgB,EAAEf,MAAMe,MAAM,CAAC,OAAO,CAAC;QAC5J,OAAOH,QAAQI,GAAG,CAAC,CAACC,IAAMC,aAAaC,IAAI,CAACF;IAC9C;IAEA,MAAMG,OAAO,AAAC,CAAA,MAAMrB,KAAK;QAAC;KAAkB,EAAE,eAAc,CAAE,CAAC,EAAE,CAACgB,MAAM;IACxE,OAAO;QAAEM,IAAI,CAAC,OAAO,EAAE9B,KAAK,CAAC,EAAEH,OAAO;QAAEgC;QAAME,UAAUpC;QAAWqC,gBAAgB,CAACvB,QAAUD,KAAKC,OAAO;QAAoBwB,YAAY,OAAOC,OAAS,AAAC,CAAA,MAAM1B,KAAK;gBAAC0B;aAAK,EAAE,eAAc,CAAE,CAAC,EAAE;IAAC;AACpM"}

package/dist/esm/embed/http.d.ts

@@ -0,0 +1,2 @@
+export declare function fetchWithRetry(url: string, init: RequestInit, baseDelayMs?: number): Promise<Response>;
+export declare function probeReachable(url: string, timeoutMs?: number): Promise<boolean>;

package/dist/esm/embed/http.js

@@ -0,0 +1,33 @@
+const MAX_ATTEMPTS = 3;
+function sleep(ms) {
+    return new Promise((resolve)=>setTimeout(resolve, ms));
+}
+// Shared retry glue for the HTTP providers: 429/5xx/network failure retry with exponential
+// backoff honoring Retry-After; any other status returns immediately for the caller to handle.
+export async function fetchWithRetry(url, init, baseDelayMs = 200) {
+    for(let attempt = 0; attempt < MAX_ATTEMPTS; attempt++){
+        try {
+            const res = await fetch(url, init);
+            if (res.ok || res.status !== 429 && res.status < 500 || attempt === MAX_ATTEMPTS - 1) return res;
+            const retryAfter = Number(res.headers.get('retry-after'));
+            // Capped: a server naming a huge Retry-After must not stall an embed run for it.
+            await sleep(retryAfter > 0 ? Math.min(retryAfter * 1000, 10000) : baseDelayMs * 2 ** attempt);
+        } catch (err) {
+            if (attempt === MAX_ATTEMPTS - 1) throw err;
+            await sleep(baseDelayMs * 2 ** attempt);
+        }
+    }
+    throw new Error('unreachable');
+}
+// Any HTTP response means the endpoint is up; only a network failure or the timeout counts as
+// unreachable, so an air-gapped machine never waits more than timeoutMs on `sense status`.
+export async function probeReachable(url, timeoutMs = 1500) {
+    try {
+        await fetch(url, {
+            signal: AbortSignal.timeout(timeoutMs)
+        });
+        return true;
+    } catch  {
+        return false;
+    }
+}

package/dist/esm/embed/http.js.map

@@ -0,0 +1 @@
+{"version":3,"sources":["/Users/kevin/Dev/OpenSource/ai/sensemaking/src/embed/http.ts"],"sourcesContent":["const MAX_ATTEMPTS = 3;\n\nfunction sleep(ms: number): Promise<void> {\n  return new Promise((resolve) => setTimeout(resolve, ms));\n}\n\n// Shared retry glue for the HTTP providers: 429/5xx/network failure retry with exponential\n// backoff honoring Retry-After; any other status returns immediately for the caller to handle.\nexport async function fetchWithRetry(url: string, init: RequestInit, baseDelayMs = 200): Promise<Response> {\n  for (let attempt = 0; attempt < MAX_ATTEMPTS; attempt++) {\n    try {\n      const res = await fetch(url, init);\n      if (res.ok || (res.status !== 429 && res.status < 500) || attempt === MAX_ATTEMPTS - 1) return res;\n      const retryAfter = Number(res.headers.get('retry-after'));\n      // Capped: a server naming a huge Retry-After must not stall an embed run for it.\n      await sleep(retryAfter > 0 ? Math.min(retryAfter * 1000, 10_000) : baseDelayMs * 2 ** attempt);\n    } catch (err) {\n      if (attempt === MAX_ATTEMPTS - 1) throw err;\n      await sleep(baseDelayMs * 2 ** attempt);\n    }\n  }\n  throw new Error('unreachable');\n}\n\n// Any HTTP response means the endpoint is up; only a network failure or the timeout counts as\n// unreachable, so an air-gapped machine never waits more than timeoutMs on `sense status`.\nexport async function probeReachable(url: string, timeoutMs = 1500): Promise<boolean> {\n  try {\n    await fetch(url, { signal: AbortSignal.timeout(timeoutMs) });\n    return true;\n  } catch {\n    return false;\n  }\n}\n"],"names":["MAX_ATTEMPTS","sleep","ms","Promise","resolve","setTimeout","fetchWithRetry","url","init","baseDelayMs","attempt","res","fetch","ok","status","retryAfter","Number","headers","get","Math","min","err","Error","probeReachable","timeoutMs","signal","AbortSignal","timeout"],"mappings":"AAAA,MAAMA,eAAe;AAErB,SAASC,MAAMC,EAAU;IACvB,OAAO,IAAIC,QAAQ,CAACC,UAAYC,WAAWD,SAASF;AACtD;AAEA,2FAA2F;AAC3F,+FAA+F;AAC/F,OAAO,eAAeI,eAAeC,GAAW,EAAEC,IAAiB,EAAEC,cAAc,GAAG;IACpF,IAAK,IAAIC,UAAU,GAAGA,UAAUV,cAAcU,UAAW;QACvD,IAAI;YACF,MAAMC,MAAM,MAAMC,MAAML,KAAKC;YAC7B,IAAIG,IAAIE,EAAE,IAAKF,IAAIG,MAAM,KAAK,OAAOH,IAAIG,MAAM,GAAG,OAAQJ,YAAYV,eAAe,GAAG,OAAOW;YAC/F,MAAMI,aAAaC,OAAOL,IAAIM,OAAO,CAACC,GAAG,CAAC;YAC1C,iFAAiF;YACjF,MAAMjB,MAAMc,aAAa,IAAII,KAAKC,GAAG,CAACL,aAAa,MAAM,SAAUN,cAAc,KAAKC;QACxF,EAAE,OAAOW,KAAK;YACZ,IAAIX,YAAYV,eAAe,GAAG,MAAMqB;YACxC,MAAMpB,MAAMQ,cAAc,KAAKC;QACjC;IACF;IACA,MAAM,IAAIY,MAAM;AAClB;AAEA,8FAA8F;AAC9F,2FAA2F;AAC3F,OAAO,eAAeC,eAAehB,GAAW,EAAEiB,YAAY,IAAI;IAChE,IAAI;QACF,MAAMZ,MAAML,KAAK;YAAEkB,QAAQC,YAAYC,OAAO,CAACH;QAAW;QAC1D,OAAO;IACT,EAAE,OAAM;QACN,OAAO;IACT;AACF"}

package/dist/esm/embed/identity.d.ts

@@ -0,0 +1,13 @@
+export declare const MODEL_FILES: string[];
+export declare const MODEL_FILENAMES: string;
+export declare function isDownloadable(model: string): boolean;
+export declare function refsMainPath(model: string): string;
+export declare function readRef(model: string): string | undefined;
+export declare function writeRef(model: string, sha: string): void;
+export declare function snapshotDir(model: string, sha: string): string;
+export declare function languagesPath(model: string): string;
+export declare function readLanguages(model: string): string[] | undefined;
+export declare function writeLanguages(model: string, languages: string[]): void;
+export declare function modelDir(model: string): string;
+export declare function hasModelFiles(model: string): boolean;
+export declare function modelIdentity(model: string): string | undefined;

package/dist/esm/embed/identity.js

@@ -0,0 +1,92 @@
+import { existsSync, mkdirSync, readFileSync, statSync, writeFileSync } from 'node:fs';
+import { homedir } from 'node:os';
+import { join } from 'node:path';
+// HF cache layout, no network: ~/.sense/models/models--<org>--<name>/{refs/main,
+// snapshots/<sha>/<files>}, files stored directly (hf_hub's blobs/symlinks layer skipped).
+export const MODEL_FILES = [
+    'model.safetensors',
+    'tokenizer.json'
+];
+// The pair, for messages that name what a local model directory must contain.
+export const MODEL_FILENAMES = MODEL_FILES.join(' and ');
+// A Hugging Face repo id: one slash, HF's charset. Anything else is a path, used as one rather
+// than resolved against the HF cache.
+const HF_MODEL_ID = /^[A-Za-z0-9][\w.-]*\/[A-Za-z0-9][\w.-]*$/;
+export function isDownloadable(model) {
+    return HF_MODEL_ID.test(model);
+}
+function repoDir(model) {
+    const [org, name] = model.split('/');
+    return join(homedir(), '.sense', 'models', `models--${org}--${name}`);
+}
+export function refsMainPath(model) {
+    return join(repoDir(model), 'refs', 'main');
+}
+// What `main` last resolved to, read from disk -- no network. Undefined until some resolution
+// (a download or an explicit prefetch) has written it once.
+export function readRef(model) {
+    const p = refsMainPath(model);
+    if (!existsSync(p)) return undefined;
+    const sha = readFileSync(p, 'utf8').trim();
+    return sha.length > 0 ? sha : undefined;
+}
+// A recorded ref pins until its snapshot directory is removed; nothing here re-resolves it.
+export function writeRef(model, sha) {
+    mkdirSync(join(repoDir(model), 'refs'), {
+        recursive: true
+    });
+    writeFileSync(refsMainPath(model), sha);
+}
+export function snapshotDir(model, sha) {
+    return join(repoDir(model), 'snapshots', sha);
+}
+// Languages describe the repo, not one weight snapshot, so this lives beside refs/, not
+// inside a snapshot dir.
+export function languagesPath(model) {
+    return join(repoDir(model), 'languages.json');
+}
+// undefined means "never resolved yet"; an empty array is a resolved fact (checked, none
+// declared) and is cached the same way, so a languageless model is not re-fetched every run.
+export function readLanguages(model) {
+    const p = languagesPath(model);
+    if (!existsSync(p)) return undefined;
+    try {
+        return JSON.parse(readFileSync(p, 'utf8'));
+    } catch  {
+        return undefined;
+    }
+}
+export function writeLanguages(model, languages) {
+    mkdirSync(repoDir(model), {
+        recursive: true
+    });
+    writeFileSync(languagesPath(model), JSON.stringify(languages));
+}
+// A local directory the caller manages, or the resolved snapshot once `main` is pinned; an HF
+// id with no recorded ref yet has no known snapshot, so this names the (fileless) repo dir.
+export function modelDir(model) {
+    if (!isDownloadable(model)) return model;
+    const sha = readRef(model);
+    return sha ? snapshotDir(model, sha) : repoDir(model);
+}
+// Both files, so an interrupted download reads as absent and the next one resumes it.
+export function hasModelFiles(model) {
+    const dir = modelDir(model);
+    return MODEL_FILES.every((file)=>existsSync(join(dir, file)));
+}
+// Weight identity for the signature, no network: a HF id's resolved sha, or the two
+// files' size+mtime for a local path. Undefined when nothing is known yet.
+export function modelIdentity(model) {
+    if (isDownloadable(model)) return readRef(model);
+    const dir = modelDir(model);
+    const parts = [];
+    for (const file of MODEL_FILES){
+        try {
+            const st = statSync(join(dir, file));
+            parts.push(`${st.size}:${Math.trunc(st.mtimeMs)}`);
+        } catch  {
+            return undefined;
+        }
+    }
+    return parts.join(',');
+}

package/dist/esm/embed/identity.js.map

@@ -0,0 +1 @@
+{"version":3,"sources":["/Users/kevin/Dev/OpenSource/ai/sensemaking/src/embed/identity.ts"],"sourcesContent":["import { existsSync, mkdirSync, readFileSync, statSync, writeFileSync } from 'node:fs';\nimport { homedir } from 'node:os';\nimport { join } from 'node:path';\n\n// HF cache layout, no network: ~/.sense/models/models--<org>--<name>/{refs/main,\n// snapshots/<sha>/<files>}, files stored directly (hf_hub's blobs/symlinks layer skipped).\n\nexport const MODEL_FILES = ['model.safetensors', 'tokenizer.json'];\n// The pair, for messages that name what a local model directory must contain.\nexport const MODEL_FILENAMES = MODEL_FILES.join(' and ');\n\n// A Hugging Face repo id: one slash, HF's charset. Anything else is a path, used as one rather\n// than resolved against the HF cache.\nconst HF_MODEL_ID = /^[A-Za-z0-9][\\w.-]*\\/[A-Za-z0-9][\\w.-]*$/;\n\nexport function isDownloadable(model: string): boolean {\n  return HF_MODEL_ID.test(model);\n}\n\nfunction repoDir(model: string): string {\n  const [org, name] = model.split('/');\n  return join(homedir(), '.sense', 'models', `models--${org}--${name}`);\n}\n\nexport function refsMainPath(model: string): string {\n  return join(repoDir(model), 'refs', 'main');\n}\n\n// What `main` last resolved to, read from disk -- no network. Undefined until some resolution\n// (a download or an explicit prefetch) has written it once.\nexport function readRef(model: string): string | undefined {\n  const p = refsMainPath(model);\n  if (!existsSync(p)) return undefined;\n  const sha = readFileSync(p, 'utf8').trim();\n  return sha.length > 0 ? sha : undefined;\n}\n\n// A recorded ref pins until its snapshot directory is removed; nothing here re-resolves it.\nexport function writeRef(model: string, sha: string): void {\n  mkdirSync(join(repoDir(model), 'refs'), { recursive: true });\n  writeFileSync(refsMainPath(model), sha);\n}\n\nexport function snapshotDir(model: string, sha: string): string {\n  return join(repoDir(model), 'snapshots', sha);\n}\n\n// Languages describe the repo, not one weight snapshot, so this lives beside refs/, not\n// inside a snapshot dir.\nexport function languagesPath(model: string): string {\n  return join(repoDir(model), 'languages.json');\n}\n\n// undefined means \"never resolved yet\"; an empty array is a resolved fact (checked, none\n// declared) and is cached the same way, so a languageless model is not re-fetched every run.\nexport function readLanguages(model: string): string[] | undefined {\n  const p = languagesPath(model);\n  if (!existsSync(p)) return undefined;\n  try {\n    return JSON.parse(readFileSync(p, 'utf8'));\n  } catch {\n    return undefined;\n  }\n}\n\nexport function writeLanguages(model: string, languages: string[]): void {\n  mkdirSync(repoDir(model), { recursive: true });\n  writeFileSync(languagesPath(model), JSON.stringify(languages));\n}\n\n// A local directory the caller manages, or the resolved snapshot once `main` is pinned; an HF\n// id with no recorded ref yet has no known snapshot, so this names the (fileless) repo dir.\nexport function modelDir(model: string): string {\n  if (!isDownloadable(model)) return model;\n  const sha = readRef(model);\n  return sha ? snapshotDir(model, sha) : repoDir(model);\n}\n\n// Both files, so an interrupted download reads as absent and the next one resumes it.\nexport function hasModelFiles(model: string): boolean {\n  const dir = modelDir(model);\n  return MODEL_FILES.every((file) => existsSync(join(dir, file)));\n}\n\n// Weight identity for the signature, no network: a HF id's resolved sha, or the two\n// files' size+mtime for a local path. Undefined when nothing is known yet.\nexport function modelIdentity(model: string): string | undefined {\n  if (isDownloadable(model)) return readRef(model);\n  const dir = modelDir(model);\n  const parts: string[] = [];\n  for (const file of MODEL_FILES) {\n    try {\n      const st = statSync(join(dir, file));\n      parts.push(`${st.size}:${Math.trunc(st.mtimeMs)}`);\n    } catch {\n      return undefined;\n    }\n  }\n  return parts.join(',');\n}\n"],"names":["existsSync","mkdirSync","readFileSync","statSync","writeFileSync","homedir","join","MODEL_FILES","MODEL_FILENAMES","HF_MODEL_ID","isDownloadable","model","test","repoDir","org","name","split","refsMainPath","readRef","p","undefined","sha","trim","length","writeRef","recursive","snapshotDir","languagesPath","readLanguages","JSON","parse","writeLanguages","languages","stringify","modelDir","hasModelFiles","dir","every","file","modelIdentity","parts","st","push","size","Math","trunc","mtimeMs"],"mappings":"AAAA,SAASA,UAAU,EAAEC,SAAS,EAAEC,YAAY,EAAEC,QAAQ,EAAEC,aAAa,QAAQ,UAAU;AACvF,SAASC,OAAO,QAAQ,UAAU;AAClC,SAASC,IAAI,QAAQ,YAAY;AAEjC,iFAAiF;AACjF,2FAA2F;AAE3F,OAAO,MAAMC,cAAc;IAAC;IAAqB;CAAiB,CAAC;AACnE,8EAA8E;AAC9E,OAAO,MAAMC,kBAAkBD,YAAYD,IAAI,CAAC,SAAS;AAEzD,+FAA+F;AAC/F,sCAAsC;AACtC,MAAMG,cAAc;AAEpB,OAAO,SAASC,eAAeC,KAAa;IAC1C,OAAOF,YAAYG,IAAI,CAACD;AAC1B;AAEA,SAASE,QAAQF,KAAa;IAC5B,MAAM,CAACG,KAAKC,KAAK,GAAGJ,MAAMK,KAAK,CAAC;IAChC,OAAOV,KAAKD,WAAW,UAAU,UAAU,CAAC,QAAQ,EAAES,IAAI,EAAE,EAAEC,MAAM;AACtE;AAEA,OAAO,SAASE,aAAaN,KAAa;IACxC,OAAOL,KAAKO,QAAQF,QAAQ,QAAQ;AACtC;AAEA,8FAA8F;AAC9F,4DAA4D;AAC5D,OAAO,SAASO,QAAQP,KAAa;IACnC,MAAMQ,IAAIF,aAAaN;IACvB,IAAI,CAACX,WAAWmB,IAAI,OAAOC;IAC3B,MAAMC,MAAMnB,aAAaiB,GAAG,QAAQG,IAAI;IACxC,OAAOD,IAAIE,MAAM,GAAG,IAAIF,MAAMD;AAChC;AAEA,4FAA4F;AAC5F,OAAO,SAASI,SAASb,KAAa,EAAEU,GAAW;IACjDpB,UAAUK,KAAKO,QAAQF,QAAQ,SAAS;QAAEc,WAAW;IAAK;IAC1DrB,cAAca,aAAaN,QAAQU;AACrC;AAEA,OAAO,SAASK,YAAYf,KAAa,EAAEU,GAAW;IACpD,OAAOf,KAAKO,QAAQF,QAAQ,aAAaU;AAC3C;AAEA,wFAAwF;AACxF,yBAAyB;AACzB,OAAO,SAASM,cAAchB,KAAa;IACzC,OAAOL,KAAKO,QAAQF,QAAQ;AAC9B;AAEA,yFAAyF;AACzF,6FAA6F;AAC7F,OAAO,SAASiB,cAAcjB,KAAa;IACzC,MAAMQ,IAAIQ,cAAchB;IACxB,IAAI,CAACX,WAAWmB,IAAI,OAAOC;IAC3B,IAAI;QACF,OAAOS,KAAKC,KAAK,CAAC5B,aAAaiB,GAAG;IACpC,EAAE,OAAM;QACN,OAAOC;IACT;AACF;AAEA,OAAO,SAASW,eAAepB,KAAa,EAAEqB,SAAmB;IAC/D/B,UAAUY,QAAQF,QAAQ;QAAEc,WAAW;IAAK;IAC5CrB,cAAcuB,cAAchB,QAAQkB,KAAKI,SAAS,CAACD;AACrD;AAEA,8FAA8F;AAC9F,4FAA4F;AAC5F,OAAO,SAASE,SAASvB,KAAa;IACpC,IAAI,CAACD,eAAeC,QAAQ,OAAOA;IACnC,MAAMU,MAAMH,QAAQP;IACpB,OAAOU,MAAMK,YAAYf,OAAOU,OAAOR,QAAQF;AACjD;AAEA,sFAAsF;AACtF,OAAO,SAASwB,cAAcxB,KAAa;IACzC,MAAMyB,MAAMF,SAASvB;IACrB,OAAOJ,YAAY8B,KAAK,CAAC,CAACC,OAAStC,WAAWM,KAAK8B,KAAKE;AAC1D;AAEA,oFAAoF;AACpF,2EAA2E;AAC3E,OAAO,SAASC,cAAc5B,KAAa;IACzC,IAAID,eAAeC,QAAQ,OAAOO,QAAQP;IAC1C,MAAMyB,MAAMF,SAASvB;IACrB,MAAM6B,QAAkB,EAAE;IAC1B,KAAK,MAAMF,QAAQ/B,YAAa;QAC9B,IAAI;YACF,MAAMkC,KAAKtC,SAASG,KAAK8B,KAAKE;YAC9BE,MAAME,IAAI,CAAC,GAAGD,GAAGE,IAAI,CAAC,CAAC,EAAEC,KAAKC,KAAK,CAACJ,GAAGK,OAAO,GAAG;QACnD,EAAE,OAAM;YACN,OAAO1B;QACT;IACF;IACA,OAAOoB,MAAMlC,IAAI,CAAC;AACpB"}

package/dist/esm/embed/langfit.d.ts

@@ -0,0 +1,6 @@
+import type { DatabaseSync } from 'node:sqlite';
+import type { EmbedProvider } from './types.js';
+type LangCounts = Record<string, number>;
+export declare function languageDistribution(db: DatabaseSync): LangCounts | undefined;
+export declare function checkLanguageFit(db: DatabaseSync, provider: EmbedProvider, texts: string[]): Promise<void>;
+export {};

package/dist/esm/embed/langfit.js

@@ -0,0 +1,44 @@
+import { getMeta, setMeta } from '../db/shared.js';
+import { SenseError } from '../errors.js';
+import { toIso3 } from './languages.js';
+const META_KEY = 'embed_languages';
+// franc's own reliability floor is a handful of characters; this is a cheap upper bound, not a
+// per-chunk minimum -- a short chunk is simply more likely to come back 'und' (unclassified).
+const SAMPLE_CHARS = 300;
+// A majority computed over a handful of chunks is noise, not a corpus signal; the decision rule
+// never fires below this many classified chunks, declared languages or not.
+const MIN_CLASSIFIED = 10;
+export function languageDistribution(db) {
+    const raw = getMeta(db, META_KEY);
+    return raw ? JSON.parse(raw) : undefined;
+}
+function classify(franc, text) {
+    const code = franc(text.slice(0, SAMPLE_CHARS));
+    return code === 'und' ? undefined : code;
+}
+// Merges chunk texts into the persisted distribution; throws (without persisting) if the model
+// declares languages and a clear majority classified so far is not among them.
+export async function checkLanguageFit(db, provider, texts) {
+    var _languageDistribution;
+    if (texts.length === 0) return;
+    const { franc } = await import('franc-min');
+    const persisted = (_languageDistribution = languageDistribution(db)) !== null && _languageDistribution !== void 0 ? _languageDistribution : {};
+    const merged = {
+        ...persisted
+    };
+    for (const text of texts){
+        var _merged_code;
+        const code = classify(franc, text);
+        if (code) merged[code] = ((_merged_code = merged[code]) !== null && _merged_code !== void 0 ? _merged_code : 0) + 1;
+    }
+    const total = Object.values(merged).reduce((a, b)=>a + b, 0);
+    if (provider.languages && provider.languages.length > 0 && total >= MIN_CLASSIFIED) {
+        const declared = new Set(provider.languages.map(toIso3));
+        const [majorityLang, majorityCount] = Object.entries(merged).sort((a, b)=>b[1] - a[1])[0];
+        if (majorityCount / total >= 0.5 && !declared.has(majorityLang)) {
+            const pct = Math.round(majorityCount / total * 100);
+            throw new SenseError('EMBED_MODEL_MISMATCH', `embed model ${provider.id} declares languages [${provider.languages.join(', ')}], but ${pct}% of this tree's classified text is "${majorityLang}" (ISO 639-3), which is not among them; choose a model for this tree's languages: \`sense init --model ...\`, see the sense-setup skill or INTEGRATIONS.md`);
+        }
+    }
+    setMeta(db, META_KEY, JSON.stringify(merged));
+}

package/dist/esm/embed/langfit.js.map

@@ -0,0 +1 @@
+{"version":3,"sources":["/Users/kevin/Dev/OpenSource/ai/sensemaking/src/embed/langfit.ts"],"sourcesContent":["import type { DatabaseSync } from 'node:sqlite';\nimport { getMeta, setMeta } from '../db/shared.ts';\nimport { SenseError } from '../errors.ts';\nimport { toIso3 } from './languages.ts';\nimport type { EmbedProvider } from './types.ts';\n\nconst META_KEY = 'embed_languages';\n// franc's own reliability floor is a handful of characters; this is a cheap upper bound, not a\n// per-chunk minimum -- a short chunk is simply more likely to come back 'und' (unclassified).\nconst SAMPLE_CHARS = 300;\n// A majority computed over a handful of chunks is noise, not a corpus signal; the decision rule\n// never fires below this many classified chunks, declared languages or not.\nconst MIN_CLASSIFIED = 10;\n\ntype LangCounts = Record<string, number>;\n\nexport function languageDistribution(db: DatabaseSync): LangCounts | undefined {\n  const raw = getMeta(db, META_KEY);\n  return raw ? JSON.parse(raw) : undefined;\n}\n\nfunction classify(franc: (text: string) => string, text: string): string | undefined {\n  const code = franc(text.slice(0, SAMPLE_CHARS));\n  return code === 'und' ? undefined : code;\n}\n\n// Merges chunk texts into the persisted distribution; throws (without persisting) if the model\n// declares languages and a clear majority classified so far is not among them.\nexport async function checkLanguageFit(db: DatabaseSync, provider: EmbedProvider, texts: string[]): Promise<void> {\n  if (texts.length === 0) return;\n  const { franc } = await import('franc-min');\n  const persisted = languageDistribution(db) ?? {};\n  const merged = { ...persisted };\n  for (const text of texts) {\n    const code = classify(franc, text);\n    if (code) merged[code] = (merged[code] ?? 0) + 1;\n  }\n\n  const total = Object.values(merged).reduce((a, b) => a + b, 0);\n  if (provider.languages && provider.languages.length > 0 && total >= MIN_CLASSIFIED) {\n    const declared = new Set(provider.languages.map(toIso3));\n    const [majorityLang, majorityCount] = Object.entries(merged).sort((a, b) => b[1] - a[1])[0];\n    if (majorityCount / total >= 0.5 && !declared.has(majorityLang)) {\n      const pct = Math.round((majorityCount / total) * 100);\n      throw new SenseError(\n        'EMBED_MODEL_MISMATCH',\n        `embed model ${provider.id} declares languages [${provider.languages.join(', ')}], but ${pct}% of this tree's classified text is \"${majorityLang}\" (ISO 639-3), which is not among them; choose a model for this tree's languages: \\`sense init --model ...\\`, see the sense-setup skill or INTEGRATIONS.md`\n      );\n    }\n  }\n  setMeta(db, META_KEY, JSON.stringify(merged));\n}\n"],"names":["getMeta","setMeta","SenseError","toIso3","META_KEY","SAMPLE_CHARS","MIN_CLASSIFIED","languageDistribution","db","raw","JSON","parse","undefined","classify","franc","text","code","slice","checkLanguageFit","provider","texts","length","persisted","merged","total","Object","values","reduce","a","b","languages","declared","Set","map","majorityLang","majorityCount","entries","sort","has","pct","Math","round","id","join","stringify"],"mappings":"AACA,SAASA,OAAO,EAAEC,OAAO,QAAQ,kBAAkB;AACnD,SAASC,UAAU,QAAQ,eAAe;AAC1C,SAASC,MAAM,QAAQ,iBAAiB;AAGxC,MAAMC,WAAW;AACjB,+FAA+F;AAC/F,8FAA8F;AAC9F,MAAMC,eAAe;AACrB,gGAAgG;AAChG,4EAA4E;AAC5E,MAAMC,iBAAiB;AAIvB,OAAO,SAASC,qBAAqBC,EAAgB;IACnD,MAAMC,MAAMT,QAAQQ,IAAIJ;IACxB,OAAOK,MAAMC,KAAKC,KAAK,CAACF,OAAOG;AACjC;AAEA,SAASC,SAASC,KAA+B,EAAEC,IAAY;IAC7D,MAAMC,OAAOF,MAAMC,KAAKE,KAAK,CAAC,GAAGZ;IACjC,OAAOW,SAAS,QAAQJ,YAAYI;AACtC;AAEA,+FAA+F;AAC/F,+EAA+E;AAC/E,OAAO,eAAeE,iBAAiBV,EAAgB,EAAEW,QAAuB,EAAEC,KAAe;QAG7Eb;IAFlB,IAAIa,MAAMC,MAAM,KAAK,GAAG;IACxB,MAAM,EAAEP,KAAK,EAAE,GAAG,MAAM,MAAM,CAAC;IAC/B,MAAMQ,aAAYf,wBAAAA,qBAAqBC,iBAArBD,mCAAAA,wBAA4B,CAAC;IAC/C,MAAMgB,SAAS;QAAE,GAAGD,SAAS;IAAC;IAC9B,KAAK,MAAMP,QAAQK,MAAO;YAEEG;QAD1B,MAAMP,OAAOH,SAASC,OAAOC;QAC7B,IAAIC,MAAMO,MAAM,CAACP,KAAK,GAAG,EAACO,eAAAA,MAAM,CAACP,KAAK,cAAZO,0BAAAA,eAAgB,KAAK;IACjD;IAEA,MAAMC,QAAQC,OAAOC,MAAM,CAACH,QAAQI,MAAM,CAAC,CAACC,GAAGC,IAAMD,IAAIC,GAAG;IAC5D,IAAIV,SAASW,SAAS,IAAIX,SAASW,SAAS,CAACT,MAAM,GAAG,KAAKG,SAASlB,gBAAgB;QAClF,MAAMyB,WAAW,IAAIC,IAAIb,SAASW,SAAS,CAACG,GAAG,CAAC9B;QAChD,MAAM,CAAC+B,cAAcC,cAAc,GAAGV,OAAOW,OAAO,CAACb,QAAQc,IAAI,CAAC,CAACT,GAAGC,IAAMA,CAAC,CAAC,EAAE,GAAGD,CAAC,CAAC,EAAE,CAAC,CAAC,EAAE;QAC3F,IAAIO,gBAAgBX,SAAS,OAAO,CAACO,SAASO,GAAG,CAACJ,eAAe;YAC/D,MAAMK,MAAMC,KAAKC,KAAK,CAAC,AAACN,gBAAgBX,QAAS;YACjD,MAAM,IAAItB,WACR,wBACA,CAAC,YAAY,EAAEiB,SAASuB,EAAE,CAAC,qBAAqB,EAAEvB,SAASW,SAAS,CAACa,IAAI,CAAC,MAAM,OAAO,EAAEJ,IAAI,qCAAqC,EAAEL,aAAa,0JAA0J,CAAC;QAEhT;IACF;IACAjC,QAAQO,IAAIJ,UAAUM,KAAKkC,SAAS,CAACrB;AACvC"}

package/dist/esm/embed/languages.d.ts

@@ -0,0 +1,3 @@
+export declare function isKnownLanguageTag(tag: string): boolean;
+export declare function toIso3(code: string): string;
+export declare const MEASURED_MODEL_LANGUAGES: Record<string, string[]>;

package/dist/esm/embed/languages.js

@@ -0,0 +1,80 @@
+// HF model cards declare languages as ISO 639-1 tags (e.g. "zh"); franc-min reports ISO 639-3
+// (e.g. "cmn"). This bridges the two for comparison. Limited to franc-min's own 82-language
+// data (README table, https://github.com/wooorm/franc), so every value here is a code franc-min
+// can actually emit; a declared tag with no entry here is compared as-is.
+const ISO_639_1_TO_3 = {
+    ar: 'arb',
+    am: 'amh',
+    az: 'azj',
+    be: 'bel',
+    bg: 'bul',
+    bn: 'ben',
+    bs: 'bos',
+    cs: 'ces',
+    de: 'deu',
+    el: 'ell',
+    en: 'eng',
+    es: 'spa',
+    fa: 'pes',
+    fr: 'fra',
+    gu: 'guj',
+    ha: 'hau',
+    hi: 'hin',
+    hr: 'hrv',
+    hu: 'hun',
+    id: 'ind',
+    ig: 'ibo',
+    it: 'ita',
+    ja: 'jpn',
+    jv: 'jav',
+    kk: 'kaz',
+    kn: 'kan',
+    ko: 'kor',
+    ml: 'mal',
+    mr: 'mar',
+    ms: 'zlm',
+    my: 'mya',
+    ne: 'npi',
+    nl: 'nld',
+    pa: 'pan',
+    pl: 'pol',
+    pt: 'por',
+    ro: 'ron',
+    ru: 'rus',
+    rw: 'kin',
+    si: 'sin',
+    so: 'som',
+    sr: 'srp',
+    su: 'sun',
+    sv: 'swe',
+    sw: 'swh',
+    ta: 'tam',
+    te: 'tel',
+    th: 'tha',
+    tl: 'tgl',
+    tr: 'tur',
+    uk: 'ukr',
+    ur: 'urd',
+    uz: 'uzn',
+    vi: 'vie',
+    yo: 'yor',
+    zh: 'cmn',
+    zu: 'zul'
+};
+export function isKnownLanguageTag(tag) {
+    return tag in ISO_639_1_TO_3;
+}
+// Normalizes a declared or detected language code to ISO 639-3 for comparison; strips a
+// region/script subtag ("zh-cn" -> "zh") first. Codes already in 639-3 form pass through.
+export function toIso3(code) {
+    var _ISO_639_1_TO_3_base;
+    const base = code.toLowerCase().split('-')[0];
+    return (_ISO_639_1_TO_3_base = ISO_639_1_TO_3[base]) !== null && _ISO_639_1_TO_3_base !== void 0 ? _ISO_639_1_TO_3_base : base;
+}
+// Languages this project measured for models whose cards declare none, so the fit check is
+// not silently off for the shipped default; asserted here because it was measured, not attested.
+export const MEASURED_MODEL_LANGUAGES = {
+    'minishlab/potion-retrieval-32M': [
+        'en'
+    ]
+};

package/dist/esm/embed/languages.js.map

@@ -0,0 +1 @@
+{"version":3,"sources":["/Users/kevin/Dev/OpenSource/ai/sensemaking/src/embed/languages.ts"],"sourcesContent":["// HF model cards declare languages as ISO 639-1 tags (e.g. \"zh\"); franc-min reports ISO 639-3\n// (e.g. \"cmn\"). This bridges the two for comparison. Limited to franc-min's own 82-language\n// data (README table, https://github.com/wooorm/franc), so every value here is a code franc-min\n// can actually emit; a declared tag with no entry here is compared as-is.\nconst ISO_639_1_TO_3: Record<string, string> = {\n  ar: 'arb',\n  am: 'amh',\n  az: 'azj',\n  be: 'bel',\n  bg: 'bul',\n  bn: 'ben',\n  bs: 'bos',\n  cs: 'ces',\n  de: 'deu',\n  el: 'ell',\n  en: 'eng',\n  es: 'spa',\n  fa: 'pes',\n  fr: 'fra',\n  gu: 'guj',\n  ha: 'hau',\n  hi: 'hin',\n  hr: 'hrv',\n  hu: 'hun',\n  id: 'ind',\n  ig: 'ibo',\n  it: 'ita',\n  ja: 'jpn',\n  jv: 'jav',\n  kk: 'kaz',\n  kn: 'kan',\n  ko: 'kor',\n  ml: 'mal',\n  mr: 'mar',\n  ms: 'zlm',\n  my: 'mya',\n  ne: 'npi',\n  nl: 'nld',\n  pa: 'pan',\n  pl: 'pol',\n  pt: 'por',\n  ro: 'ron',\n  ru: 'rus',\n  rw: 'kin',\n  si: 'sin',\n  so: 'som',\n  sr: 'srp',\n  su: 'sun',\n  sv: 'swe',\n  sw: 'swh',\n  ta: 'tam',\n  te: 'tel',\n  th: 'tha',\n  tl: 'tgl',\n  tr: 'tur',\n  uk: 'ukr',\n  ur: 'urd',\n  uz: 'uzn',\n  vi: 'vie',\n  yo: 'yor',\n  zh: 'cmn',\n  zu: 'zul',\n};\n\nexport function isKnownLanguageTag(tag: string): boolean {\n  return tag in ISO_639_1_TO_3;\n}\n\n// Normalizes a declared or detected language code to ISO 639-3 for comparison; strips a\n// region/script subtag (\"zh-cn\" -> \"zh\") first. Codes already in 639-3 form pass through.\nexport function toIso3(code: string): string {\n  const base = code.toLowerCase().split('-')[0];\n  return ISO_639_1_TO_3[base] ?? base;\n}\n\n// Languages this project measured for models whose cards declare none, so the fit check is\n// not silently off for the shipped default; asserted here because it was measured, not attested.\nexport const MEASURED_MODEL_LANGUAGES: Record<string, string[]> = {\n  'minishlab/potion-retrieval-32M': ['en'], // BENCHMARKING.md bake-off; token-loss measurements\n};\n"],"names":["ISO_639_1_TO_3","ar","am","az","be","bg","bn","bs","cs","de","el","en","es","fa","fr","gu","ha","hi","hr","hu","id","ig","it","ja","jv","kk","kn","ko","ml","mr","ms","my","ne","nl","pa","pl","pt","ro","ru","rw","si","so","sr","su","sv","sw","ta","te","th","tl","tr","uk","ur","uz","vi","yo","zh","zu","isKnownLanguageTag","tag","toIso3","code","base","toLowerCase","split","MEASURED_MODEL_LANGUAGES"],"mappings":"AAAA,8FAA8F;AAC9F,4FAA4F;AAC5F,gGAAgG;AAChG,0EAA0E;AAC1E,MAAMA,iBAAyC;IAC7CC,IAAI;IACJC,IAAI;IACJC,IAAI;IACJC,IAAI;IACJC,IAAI;IACJC,IAAI;IACJC,IAAI;IACJC,IAAI;IACJC,IAAI;IACJC,IAAI;IACJC,IAAI;IACJC,IAAI;IACJC,IAAI;IACJC,IAAI;IACJC,IAAI;IACJC,IAAI;IACJC,IAAI;IACJC,IAAI;IACJC,IAAI;IACJC,IAAI;IACJC,IAAI;IACJC,IAAI;IACJC,IAAI;IACJC,IAAI;IACJC,IAAI;IACJC,IAAI;IACJC,IAAI;IACJC,IAAI;IACJC,IAAI;IACJC,IAAI;IACJC,IAAI;IACJC,IAAI;IACJC,IAAI;IACJC,IAAI;IACJC,IAAI;IACJC,IAAI;IACJC,IAAI;IACJC,IAAI;IACJC,IAAI;IACJC,IAAI;IACJC,IAAI;IACJC,IAAI;IACJC,IAAI;IACJC,IAAI;IACJC,IAAI;IACJC,IAAI;IACJC,IAAI;IACJC,IAAI;IACJC,IAAI;IACJC,IAAI;IACJC,IAAI;IACJC,IAAI;IACJC,IAAI;IACJC,IAAI;IACJC,IAAI;IACJC,IAAI;IACJC,IAAI;AACN;AAEA,OAAO,SAASC,mBAAmBC,GAAW;IAC5C,OAAOA,OAAO3D;AAChB;AAEA,wFAAwF;AACxF,0FAA0F;AAC1F,OAAO,SAAS4D,OAAOC,IAAY;QAE1B7D;IADP,MAAM8D,OAAOD,KAAKE,WAAW,GAAGC,KAAK,CAAC,IAAI,CAAC,EAAE;IAC7C,QAAOhE,uBAAAA,cAAc,CAAC8D,KAAK,cAApB9D,kCAAAA,uBAAwB8D;AACjC;AAEA,2FAA2F;AAC3F,iGAAiG;AACjG,OAAO,MAAMG,2BAAqD;IAChE,kCAAkC;QAAC;KAAK;AAC1C,EAAE"}