sensemaking 0.18.3 → 0.19.0

package/dist/esm/config/types.js.map

@@ -1 +1 @@
-{"version":3,"sources":["/Users/kevin/Dev/OpenSource/ai/sensemaking/src/config/types.ts"],"sourcesContent":["import type { SignalWeights } from './signals.ts';\n\nexport const CONFIG_FILENAME = 'sense.config.json';\nexport const STATE_DIR = '.sense';\n\n// Highest sense.config.json `version` this build understands. Older versions auto-migrate on load.\nexport const SUPPORTED_CONFIG_VERSION = 5;\n\n// Each feature owns its tables, parse-time extraction, and reconcile step; commands degrade when one is off.\n// links/sections/tags/rank are opt-out toggles in the top-level `features` block; embed is not a\n// member of that block -- it is on exactly when an `embed` block names a model.\nexport const FEATURE_NAMES = ['links', 'sections', 'tags', 'rank', 'embed'] as const;\nexport type FeatureName = (typeof FEATURE_NAMES)[number];\n\n// `embed` names the model and gives the tree vectors at all; a preset's `signals` says which\n// engines that scope's searches use.\nexport interface EmbedConfig {\n  model?: string;\n  provider?: 'static' | 'openai' | 'cohere';\n  url?: string; // openai provider: OpenAI-compatible base URL, e.g. http://localhost:11434/v1\n  key?: string; // openai/cohere provider: name of the env var holding the bearer token, if any\n  languages?: string[]; // owner-declared model languages; enables the language-fit check where no card exists\n  chunkTokens?: number; // chunk size ceiling in estimated tokens for small-context models; default 500\n}\n\nexport const DEFAULT_EMBED_MODEL = 'minishlab/potion-retrieval-32M';\n\n// The single declaration every other store-name union derives from (Store.name in\n// store/types.ts, the registry in store/index.ts): one store gets added here, not in four\n// hand-maintained copies.\nexport type StoreName = 'sqlite' | 'duckdb' | 'turso';\n\n// A named, self-contained file-selection scope. include/exclude are globby patterns resolved\n// relative to the config file. No inheritance between presets -- every field a preset wants\n// it declares itself. Presets may overlap freely; they are views, not partitions.\nexport interface Preset {\n  include: string[];\n  exclude?: string[];\n  k?: number; // result count for `search` scoped to this preset; default 10\n  // Exhaustive when present: every signal this preset's searches compose, each mapped to\n  // its RRF weight (a finite number > 0; presence is enablement). Absent means every signal\n  // whose prerequisite holds, each at weight 1 -- words always, links when the feature is on,\n  // vectors when the tree names a model. A layer searched for exact wording (ingested sources,\n  // archives, generated output) declares {\"words\":1,\"links\":1} and costs no embedding at all.\n  signals?: SignalWeights;\n  where?: string; // standing SQL condition against frontmatter alias `f`\n}\n\n// A saved `search` invocation: `sense <name>` runs like\n// `sense search <search> [--preset] [--include] [--exclude] [--where] [--k]`.\nexport interface SavedSearch {\n  search: string;\n  preset?: string;\n  include?: string[];\n  exclude?: string[];\n  where?: string;\n  k?: number;\n}\n\n// An entry names the verb it runs, one to one with the commands: { sql } runs like\n// `sense sql`, { search } like `sense search`.\nexport type SavedQuery = { sql: string } | SavedSearch;\n\nexport interface Config {\n  // Editor-only pointer to schema.json; never read by sense.\n  $schema?: string;\n  version?: number;\n  // File selection, index-time, and per-preset search defaults. A file is indexed iff any\n  // preset's include/exclude covers it (union). `default` is used when a command names no preset.\n  presets: Record<string, Preset>;\n  // Global feature defaults; absent block or key means enabled. embed is not a member here --\n  // see the `embed` block.\n  features?: { links?: boolean; sections?: boolean; tags?: boolean; rank?: boolean };\n  // Names the embedding model. Present means every indexed file gets vectors; absent means none do.\n  embed?: EmbedConfig;\n  // Settings for the `content` FTS5 table. `tokenize` decides which languages the tree can be\n  // word-searched in at all; changing it rebuilds the index.\n  content?: { tokenize?: string };\n  // Backing store engine for this tree: \"sqlite\" (default, zero-dependency), \"duckdb\", or\n  // \"turso\" (each an optional dependency, installed on first use). All three implement the same\n  // commands; \"turso\" implements only the portable surface so far (no lexical, no vectors).\n  store?: StoreName;\n  queries: Record<string, SavedQuery>;\n}\n\nexport interface ResolvedConfig extends Config {\n  baseDir: string;\n  configPath: string | null;\n  // Keys the file declares that this build does not read; cli reports them.\n  unknownKeys?: string[];\n  // Set when loadConfig auto-migrated the file on disk; cli reports it.\n  migratedFrom?: number;\n}\n\nexport interface SearchOverrides {\n  preset?: string;\n  k?: number;\n  where?: string;\n  include?: string[];\n  exclude?: string[];\n  noExclude?: boolean; // drop the preset's exclude for this command; --exclude still applies\n}\n\nexport interface EffectiveSearch {\n  presetName: string;\n  k: number;\n  where?: string;\n  include: string[];\n  exclude?: string[];\n  signals: SignalWeights; // this preset's effective signal weights\n}\n"],"names":["CONFIG_FILENAME","STATE_DIR","SUPPORTED_CONFIG_VERSION","FEATURE_NAMES","DEFAULT_EMBED_MODEL"],"mappings":"AAEA,OAAO,MAAMA,kBAAkB,oBAAoB;AACnD,OAAO,MAAMC,YAAY,SAAS;AAElC,mGAAmG;AACnG,OAAO,MAAMC,2BAA2B,EAAE;AAE1C,6GAA6G;AAC7G,iGAAiG;AACjG,gFAAgF;AAChF,OAAO,MAAMC,gBAAgB;IAAC;IAAS;IAAY;IAAQ;IAAQ;CAAQ,CAAU;AAcrF,OAAO,MAAMC,sBAAsB,iCAAiC"}
+{"version":3,"sources":["/Users/kevin/Dev/OpenSource/ai/sensemaking/src/config/types.ts"],"sourcesContent":["import type { SignalWeights } from './signals.ts';\n\nexport const CONFIG_FILENAME = 'sense.config.json';\nexport const STATE_DIR = '.sense';\n\n// Highest sense.config.json `version` this build understands. Older versions auto-migrate on load.\nexport const SUPPORTED_CONFIG_VERSION = 5;\n\n// Each feature owns its tables, parse-time extraction, and reconcile step; commands degrade when one is off.\n// links/sections/tags/rank are opt-out toggles in `features`; embed is on exactly when an `embed` block names a model.\nexport const FEATURE_NAMES = ['links', 'sections', 'tags', 'rank', 'embed'] as const;\nexport type FeatureName = (typeof FEATURE_NAMES)[number];\n\n// `embed` names the model and gives the tree vectors at all; a preset's `signals` says which\n// engines that scope's searches use.\nexport interface EmbedConfig {\n  model?: string;\n  provider?: 'static' | 'openai' | 'cohere';\n  url?: string; // openai provider: OpenAI-compatible base URL, e.g. http://localhost:11434/v1\n  key?: string; // openai/cohere provider: name of the env var holding the bearer token, if any\n  languages?: string[]; // owner-declared model languages; enables the language-fit check where no card exists\n  chunkTokens?: number; // chunk size ceiling in estimated tokens for small-context models; default 500\n}\n\nexport const DEFAULT_EMBED_MODEL = 'minishlab/potion-retrieval-32M';\n\n// The single declaration every other store name derives from (Store.name, the store/index.ts\n// registry, validate.ts's accepted set). An array, not a bare union, so runtime checks read the same list.\nexport const STORE_NAMES = ['sqlite', 'duckdb', 'turso'] as const;\nexport type StoreName = (typeof STORE_NAMES)[number];\n\n// A named, self-contained file-selection scope: include/exclude are globby patterns resolved\n// relative to the config file. No inheritance between presets; they may overlap freely (views, not partitions).\nexport interface Preset {\n  include: string[];\n  exclude?: string[];\n  k?: number; // result count for `search` scoped to this preset; default 10\n  // Exhaustive when present: every signal this preset's searches compose, each mapped to its RRF\n  // weight (finite > 0; presence is enablement). Absent = every signal whose prerequisite holds, at weight 1 (words always, links when on, vectors when a model is named).\n  signals?: SignalWeights;\n  where?: string; // standing SQL condition against frontmatter alias `f`\n}\n\n// A saved `search` invocation: `sense <name>` runs like\n// `sense search <search> [--preset] [--include] [--exclude] [--where] [--k]`.\nexport interface SavedSearch {\n  search: string;\n  preset?: string;\n  include?: string[];\n  exclude?: string[];\n  where?: string;\n  k?: number;\n}\n\n// An entry names the verb it runs, one to one with the commands: { sql } runs like\n// `sense sql`, { search } like `sense search`.\nexport type SavedQuery = { sql: string } | SavedSearch;\n\nexport interface Config {\n  // Editor-only pointer to schema.json; never read by sense.\n  $schema?: string;\n  version?: number;\n  // File selection, index-time, and per-preset search defaults. A file is indexed iff any\n  // preset's include/exclude covers it (union). `default` is used when a command names no preset.\n  presets: Record<string, Preset>;\n  // Global feature defaults; absent block or key means enabled. embed is not a member here --\n  // see the `embed` block.\n  features?: { links?: boolean; sections?: boolean; tags?: boolean; rank?: boolean };\n  // Names the embedding model. Present means every indexed file gets vectors; absent means none do.\n  embed?: EmbedConfig;\n  // Settings for the `content` FTS5 table. `tokenize` decides which languages the tree can be\n  // word-searched in at all; changing it rebuilds the index.\n  content?: { tokenize?: string };\n  // Backing store engine: \"sqlite\" (default, zero-dependency), \"duckdb\", or \"turso\" (optional\n  // dependency, installed on first use). What differs per store is declared through CAPABILITIES (store/types.ts).\n  store?: StoreName;\n  queries: Record<string, SavedQuery>;\n}\n\nexport interface ResolvedConfig extends Config {\n  baseDir: string;\n  configPath: string | null;\n  // Keys the file declares that this build does not read; cli reports them.\n  unknownKeys?: string[];\n  // Set when loadConfig auto-migrated the file on disk; cli reports it.\n  migratedFrom?: number;\n}\n\nexport interface SearchOverrides {\n  preset?: string;\n  k?: number;\n  where?: string;\n  include?: string[];\n  exclude?: string[];\n  noExclude?: boolean; // drop the preset's exclude for this command; --exclude still applies\n}\n\nexport interface EffectiveSearch {\n  presetName: string;\n  k: number;\n  where?: string;\n  include: string[];\n  exclude?: string[];\n  signals: SignalWeights; // this preset's effective signal weights\n}\n"],"names":["CONFIG_FILENAME","STATE_DIR","SUPPORTED_CONFIG_VERSION","FEATURE_NAMES","DEFAULT_EMBED_MODEL","STORE_NAMES"],"mappings":"AAEA,OAAO,MAAMA,kBAAkB,oBAAoB;AACnD,OAAO,MAAMC,YAAY,SAAS;AAElC,mGAAmG;AACnG,OAAO,MAAMC,2BAA2B,EAAE;AAE1C,6GAA6G;AAC7G,uHAAuH;AACvH,OAAO,MAAMC,gBAAgB;IAAC;IAAS;IAAY;IAAQ;IAAQ;CAAQ,CAAU;AAcrF,OAAO,MAAMC,sBAAsB,iCAAiC;AAEpE,6FAA6F;AAC7F,2GAA2G;AAC3G,OAAO,MAAMC,cAAc;IAAC;IAAU;IAAU;CAAQ,CAAU"}

package/dist/esm/config/validate.d.ts

@@ -1,4 +1,4 @@
-import type { Config } from './types.js';
+import { type Config } from './types.js';
 export declare const KNOWN_EMBED_KEYS: Set<string>;
 export declare function unknownConfigKeys(cfg: Record<string, unknown>): string[];
 export declare function validateLegacyScan(parsed: unknown, configPath: string): void;

package/dist/esm/config/validate.js

@@ -1,5 +1,6 @@
 import { SenseError } from '../errors.js';
 import { SIGNAL_NAMES, SIGNAL_PREREQUISITES } from './signals.js';
+import { STORE_NAMES } from './types.js';
 // Shape check for hand-edited files, so a typo names itself instead of surfacing as a
 // TypeError. Unknown top-level keys warn (forward compat); unknown keys inside a block error.
 const KNOWN_KEYS = new Set([
@@ -76,12 +77,9 @@ function validateContentBlock(value, configPath) {
         throw new SenseError('CONFIG_INVALID', `${configPath}: content.tokenize must be a non-empty string, e.g. "trigram" or "unicode61 tokenchars '-_'"`);
     }
 }
-// "turso" is implemented but not offered: it has no lexical or vector search yet, so a tree
-// naming it could not search. Registered in the store registry and exercised by the parity
-// suite; re-add here and in schema.json's enum when those land.
 function validateStoreKey(value, configPath) {
-    if (value !== 'sqlite' && value !== 'duckdb') {
-        throw new SenseError('CONFIG_INVALID', `${configPath}: store must be "sqlite" or "duckdb"`);
+    if (!STORE_NAMES.includes(value)) {
+        throw new SenseError('CONFIG_INVALID', `${configPath}: store must be ${STORE_NAMES.map((name)=>`"${name}"`).join(' or ')}`);
     }
 }
 function validateFeaturesBlock(value, configPath) {
@@ -143,9 +141,8 @@ function isSignalName(value) {
 function isSignalWeight(value) {
     return typeof value === 'number' && Number.isFinite(value) && value > 0;
 }
-// Shape (an object of signal name -> weight) plus the one cross-signal prerequisite that
-// doesn't need the top-level embed block: links is seeded by word-match rows, so declaring it
-// without words is always empty.
+// Shape (an object of signal name -> weight) plus the one cross-signal prerequisite that doesn't
+// need the top-level embed block: links is seeded by word-match rows, so declaring it without words is always empty.
 function validatePresetSignals(name, value, configPath) {
     if (typeof value !== 'object' || value === null || Array.isArray(value) || Object.keys(value).length === 0) {
         throw new SenseError('CONFIG_INVALID', `${configPath}: presets.${name}.signals must be a non-empty object of signal name -> weight (a finite number > 0), naming ${SIGNAL_NAMES.join(', ')} -- a preset that searches nothing is a mistake; name at least "words", or omit "signals" to use every signal whose prerequisites hold at weight 1`);
@@ -192,8 +189,8 @@ function validatePreset(name, value, configPath) {
 }
 // A queries.<name> entry: { sql } or a saved search { search, preset?, include?, exclude?, where?, k? }.
 function validateSavedQuery(name, value, configPath) {
-    // A bare string used to mean SQL. It now fails rather than being inferred: an entry says
-    // which of the two verbs it runs, the same choice the CLI makes explicit.
+    // A bare string is rejected rather than inferred to be SQL: an entry says which of the two
+    // verbs it runs, the same choice the CLI makes explicit.
     if (typeof value === 'string') {
         throw new SenseError('CONFIG_INVALID', `${configPath}: queries.${name} must say which verb it runs: { "sql": ${JSON.stringify(value.length > 40 ? `${value.slice(0, 37)}...` : value)} } to run it as SQL, or { "search": "..." } for a ranked search`);
     }

package/dist/esm/config/validate.js.map

@@ -1 +1 @@
-{"version":3,"sources":["/Users/kevin/Dev/OpenSource/ai/sensemaking/src/config/validate.ts"],"sourcesContent":["import { SenseError } from '../errors.ts';\nimport { SIGNAL_NAMES, SIGNAL_PREREQUISITES, type SignalName } from './signals.ts';\nimport type { Config } from './types.ts';\n\n// Shape check for hand-edited files, so a typo names itself instead of surfacing as a\n// TypeError. Unknown top-level keys warn (forward compat); unknown keys inside a block error.\nconst KNOWN_KEYS = new Set(['$schema', 'version', 'presets', 'features', 'embed', 'content', 'store', 'queries']);\nconst KNOWN_PRESET_KEYS = new Set(['include', 'exclude', 'k', 'signals', 'where']);\nconst KNOWN_FEATURE_KEYS = new Set(['links', 'sections', 'tags', 'rank']);\n// Exported so a docs test can assert every key here has a schema.json property -- one source\n// of truth for what the embed block accepts, checked against the other for drift.\nexport const KNOWN_EMBED_KEYS = new Set(['model', 'provider', 'url', 'key', 'languages', 'chunkTokens']);\nconst KNOWN_CONTENT_KEYS = new Set(['tokenize']);\nconst SAVED_SEARCH_KEYS = new Set(['search', 'preset', 'include', 'exclude', 'where', 'k']);\n\nexport function unknownConfigKeys(cfg: Record<string, unknown>): string[] {\n  return Object.keys(cfg).filter((k) => !KNOWN_KEYS.has(k));\n}\n\nfunction isNonEmptyStringArray(value: unknown): value is string[] {\n  return Array.isArray(value) && value.length > 0 && value.every((g) => typeof g === 'string' && g.length > 0);\n}\n\n// Pre-v3 shape check, just enough for migrateConfig to run safely (it reads scan.include\n// directly). The full v3 shape is checked by validateConfig once migration has produced it.\nexport function validateLegacyScan(parsed: unknown, configPath: string): void {\n  const cfg = typeof parsed === 'object' && parsed !== null && !Array.isArray(parsed) ? (parsed as Record<string, unknown>) : null;\n  const scan = cfg?.scan as { include?: unknown } | undefined;\n  if (!cfg || !scan || !isNonEmptyStringArray(scan.include)) {\n    throw new SenseError('CONFIG_INVALID', `${configPath}: scan.include must be a non-empty array of glob strings`);\n  }\n}\n\n// Only the shape is checked here. Whether the linked SQLite accepts the tokenizer is settled\n// by probing it in store/sqlite/open.ts, so this never has to carry a table of which version added what.\nfunction validateContentBlock(value: unknown, configPath: string): void {\n  if (typeof value !== 'object' || value === null || Array.isArray(value)) {\n    throw new SenseError('CONFIG_INVALID', `${configPath}: content must be an object of { tokenize?: string }`);\n  }\n  const block = value as Record<string, unknown>;\n  const unknown = Object.keys(block).filter((k) => !KNOWN_CONTENT_KEYS.has(k));\n  if (unknown.length > 0) {\n    throw new SenseError('CONFIG_INVALID', `${configPath}: content has unknown key(s) ${unknown.join(', ')}; content takes tokenize`);\n  }\n  if (block.tokenize !== undefined && (typeof block.tokenize !== 'string' || block.tokenize.trim() === '')) {\n    throw new SenseError('CONFIG_INVALID', `${configPath}: content.tokenize must be a non-empty string, e.g. \"trigram\" or \"unicode61 tokenchars '-_'\"`);\n  }\n}\n\n// \"turso\" is implemented but not offered: it has no lexical or vector search yet, so a tree\n// naming it could not search. Registered in the store registry and exercised by the parity\n// suite; re-add here and in schema.json's enum when those land.\nfunction validateStoreKey(value: unknown, configPath: string): void {\n  if (value !== 'sqlite' && value !== 'duckdb') {\n    throw new SenseError('CONFIG_INVALID', `${configPath}: store must be \"sqlite\" or \"duckdb\"`);\n  }\n}\n\nfunction validateFeaturesBlock(value: unknown, configPath: string): void {\n  if (typeof value !== 'object' || value === null || Array.isArray(value)) {\n    throw new SenseError('CONFIG_INVALID', `${configPath}: features must be an object of name -> boolean`);\n  }\n  const block = value as Record<string, unknown>;\n  const unknown = Object.keys(block).filter((k) => !KNOWN_FEATURE_KEYS.has(k));\n  if (unknown.length > 0) {\n    throw new SenseError('CONFIG_INVALID', `${configPath}: features has unknown key(s) ${unknown.join(', ')}; embed is not a features key -- see the top-level \"embed\" block`);\n  }\n  for (const [name, v] of Object.entries(block)) {\n    if (typeof v !== 'boolean') {\n      throw new SenseError('CONFIG_INVALID', `${configPath}: features.${name} must be a boolean`);\n    }\n  }\n}\n\nfunction validateEmbedBlock(value: unknown, configPath: string): void {\n  if (typeof value !== 'object' || value === null || Array.isArray(value)) {\n    throw new SenseError('CONFIG_INVALID', `${configPath}: embed must be an object of { model?, provider?: \"static\"|\"openai\"|\"cohere\", url?, key? }`);\n  }\n  const embed = value as Record<string, unknown>;\n  const unknown = Object.keys(embed).filter((k) => !KNOWN_EMBED_KEYS.has(k));\n  if (unknown.length > 0) {\n    throw new SenseError('CONFIG_INVALID', `${configPath}: embed has unknown key(s) ${unknown.join(', ')}`);\n  }\n  if (embed.model !== undefined && typeof embed.model !== 'string') {\n    throw new SenseError('CONFIG_INVALID', `${configPath}: embed.model must be a string`);\n  }\n  // The block exists to name a model; provider settings without one would read as \"vectors\n  // configured\" while embedding stays off, which is the ambiguity v4 removed.\n  if (embed.model === undefined || embed.model === '') {\n    throw new SenseError('CONFIG_INVALID', `${configPath}: embed.model is required when the \"embed\" block is present (it is what turns vectors on); remove the block to index without vectors`);\n  }\n  if (embed.provider !== undefined && embed.provider !== 'static' && embed.provider !== 'openai' && embed.provider !== 'cohere') {\n    throw new SenseError('CONFIG_INVALID', `${configPath}: embed.provider must be \"static\", \"openai\", or \"cohere\"`);\n  }\n  if (embed.url !== undefined && typeof embed.url !== 'string') {\n    throw new SenseError('CONFIG_INVALID', `${configPath}: embed.url must be a string`);\n  }\n  if (embed.key !== undefined && typeof embed.key !== 'string') {\n    throw new SenseError('CONFIG_INVALID', `${configPath}: embed.key must be a string`);\n  }\n  // openai has no default host, so a missing url would otherwise only surface at the first\n  // vector search; a config error belongs at config time.\n  if (embed.provider === 'openai' && (embed.url === undefined || embed.url === '')) {\n    throw new SenseError('CONFIG_INVALID', `${configPath}: embed.provider \"openai\" requires embed.url (e.g. \"http://localhost:11434/v1\" for Ollama)`);\n  }\n  if (embed.languages !== undefined && (!Array.isArray(embed.languages) || embed.languages.length === 0 || embed.languages.some((l) => typeof l !== 'string' || l === ''))) {\n    throw new SenseError('CONFIG_INVALID', `${configPath}: embed.languages must be a non-empty array of language tags (e.g. [\"en\", \"zh\"])`);\n  }\n  if (embed.chunkTokens !== undefined && (typeof embed.chunkTokens !== 'number' || !Number.isInteger(embed.chunkTokens) || embed.chunkTokens <= 0)) {\n    throw new SenseError('CONFIG_INVALID', `${configPath}: embed.chunkTokens must be a positive integer`);\n  }\n}\n\nfunction isSignalName(value: unknown): value is SignalName {\n  return typeof value === 'string' && (SIGNAL_NAMES as readonly string[]).includes(value);\n}\n\nfunction isSignalWeight(value: unknown): value is number {\n  return typeof value === 'number' && Number.isFinite(value) && value > 0;\n}\n\n// Shape (an object of signal name -> weight) plus the one cross-signal prerequisite that\n// doesn't need the top-level embed block: links is seeded by word-match rows, so declaring it\n// without words is always empty.\nfunction validatePresetSignals(name: string, value: unknown, configPath: string): void {\n  if (typeof value !== 'object' || value === null || Array.isArray(value) || Object.keys(value).length === 0) {\n    throw new SenseError(\n      'CONFIG_INVALID',\n      `${configPath}: presets.${name}.signals must be a non-empty object of signal name -> weight (a finite number > 0), naming ${SIGNAL_NAMES.join(', ')} -- a preset that searches nothing is a mistake; name at least \"words\", or omit \"signals\" to use every signal whose prerequisites hold at weight 1`\n    );\n  }\n  const entries = Object.entries(value as Record<string, unknown>);\n  const unknown = entries.filter(([k]) => !isSignalName(k)).map(([k]) => k);\n  if (unknown.length > 0) {\n    throw new SenseError('CONFIG_INVALID', `${configPath}: presets.${name}.signals names unknown signal(s) ${unknown.join(', ')}; valid signals are ${SIGNAL_NAMES.join(', ')}`);\n  }\n  for (const [signal, weight] of entries) {\n    if (!isSignalWeight(weight)) {\n      throw new SenseError('CONFIG_INVALID', `${configPath}: presets.${name}.signals.${signal} must be a finite number > 0`);\n    }\n  }\n  const names = new Set(entries.map(([k]) => k));\n  for (const [signal, requires] of Object.entries(SIGNAL_PREREQUISITES) as Array<[SignalName, SignalName]>) {\n    if (names.has(signal) && !names.has(requires)) {\n      throw new SenseError('CONFIG_INVALID', `${configPath}: presets.${name}.signals has \"${signal}\" without \"${requires}\"; add \"${requires}\" to presets.${name}.signals, or drop \"${signal}\"`);\n    }\n  }\n}\n\nfunction validatePreset(name: string, value: unknown, configPath: string): void {\n  if (typeof value !== 'object' || value === null || Array.isArray(value)) {\n    throw new SenseError('CONFIG_INVALID', `${configPath}: presets.${name} must be an object`);\n  }\n  const preset = value as Record<string, unknown>;\n  const unknown = Object.keys(preset).filter((k) => !KNOWN_PRESET_KEYS.has(k));\n  if (unknown.length > 0) {\n    throw new SenseError('CONFIG_INVALID', `${configPath}: presets.${name} has unknown key(s) ${unknown.join(', ')}; a preset takes include, exclude, k, signals, where`);\n  }\n  if (!isNonEmptyStringArray(preset.include)) {\n    throw new SenseError('CONFIG_INVALID', `${configPath}: presets.${name}.include must be a non-empty array of glob strings`);\n  }\n  if (preset.exclude !== undefined && !isNonEmptyStringArray(preset.exclude)) {\n    throw new SenseError('CONFIG_INVALID', `${configPath}: presets.${name}.exclude must be a non-empty array of glob strings`);\n  }\n  if (preset.signals !== undefined) validatePresetSignals(name, preset.signals, configPath);\n  if (preset.k !== undefined && (typeof preset.k !== 'number' || !Number.isInteger(preset.k) || preset.k <= 0)) {\n    throw new SenseError('CONFIG_INVALID', `${configPath}: presets.${name}.k must be a positive integer`);\n  }\n  if (preset.where !== undefined && typeof preset.where !== 'string') {\n    throw new SenseError('CONFIG_INVALID', `${configPath}: presets.${name}.where must be a SQL condition string`);\n  }\n}\n\n// A queries.<name> entry: { sql } or a saved search { search, preset?, include?, exclude?, where?, k? }.\nfunction validateSavedQuery(name: string, value: unknown, configPath: string): void {\n  // A bare string used to mean SQL. It now fails rather than being inferred: an entry says\n  // which of the two verbs it runs, the same choice the CLI makes explicit.\n  if (typeof value === 'string') {\n    throw new SenseError('CONFIG_INVALID', `${configPath}: queries.${name} must say which verb it runs: { \"sql\": ${JSON.stringify(value.length > 40 ? `${value.slice(0, 37)}...` : value)} } to run it as SQL, or { \"search\": \"...\" } for a ranked search`);\n  }\n  if (typeof value !== 'object' || value === null || Array.isArray(value)) {\n    throw new SenseError('CONFIG_INVALID', `${configPath}: queries.${name} must be { sql } or { search, preset?, include?, exclude?, where?, k? }`);\n  }\n  const entry = value as Record<string, unknown>;\n  if ('sql' in entry) {\n    const unknown = Object.keys(entry).filter((k) => k !== 'sql');\n    if (unknown.length > 0) {\n      throw new SenseError('CONFIG_INVALID', `${configPath}: queries.${name} has unknown key(s) ${unknown.join(', ')}; a { sql } query takes only sql`);\n    }\n    if (typeof entry.sql !== 'string' || entry.sql.trim() === '') {\n      throw new SenseError('CONFIG_INVALID', `${configPath}: queries.${name}.sql must be a non-empty string`);\n    }\n    return;\n  }\n  if ('search' in entry) {\n    const unknown = Object.keys(entry).filter((k) => !SAVED_SEARCH_KEYS.has(k));\n    if (unknown.length > 0) {\n      throw new SenseError('CONFIG_INVALID', `${configPath}: queries.${name} has unknown key(s) ${unknown.join(', ')}; a saved search takes search, preset, include, exclude, where, k`);\n    }\n    // A saved entry saves a question; a scope without a question is just flags.\n    if (typeof entry.search !== 'string' || entry.search.trim() === '') {\n      throw new SenseError('CONFIG_INVALID', `${configPath}: queries.${name}.search must be non-empty text`);\n    }\n    if (entry.preset !== undefined && (typeof entry.preset !== 'string' || entry.preset.length === 0)) {\n      throw new SenseError('CONFIG_INVALID', `${configPath}: queries.${name}.preset must be a preset name`);\n    }\n    if (entry.include !== undefined && !isNonEmptyStringArray(entry.include)) {\n      throw new SenseError('CONFIG_INVALID', `${configPath}: queries.${name}.include must be a non-empty array of glob strings`);\n    }\n    if (entry.exclude !== undefined && !isNonEmptyStringArray(entry.exclude)) {\n      throw new SenseError('CONFIG_INVALID', `${configPath}: queries.${name}.exclude must be a non-empty array of glob strings`);\n    }\n    if (entry.where !== undefined && typeof entry.where !== 'string') {\n      throw new SenseError('CONFIG_INVALID', `${configPath}: queries.${name}.where must be a SQL condition string`);\n    }\n    if (entry.k !== undefined && (typeof entry.k !== 'number' || !Number.isInteger(entry.k) || entry.k <= 0)) {\n      throw new SenseError('CONFIG_INVALID', `${configPath}: queries.${name}.k must be a positive integer`);\n    }\n    return;\n  }\n  throw new SenseError('CONFIG_INVALID', `${configPath}: queries.${name} must be { sql } or { search, preset?, include?, exclude?, where?, k? }`);\n}\n\nexport function validateConfig(parsed: unknown, configPath: string): Config {\n  if (typeof parsed !== 'object' || parsed === null || Array.isArray(parsed)) {\n    throw new SenseError('CONFIG_INVALID', `${configPath}: config must be a JSON object`);\n  }\n  const cfg = parsed as Record<string, unknown>;\n\n  // `checks` is rejected by name, not warned: silence would hide that assertions are gone.\n  if (cfg.checks !== undefined) {\n    throw new SenseError('CONFIG_INVALID', `${configPath}: checks was removed in v3 -- sense check no longer asserts on saved queries; a returned row set is the reader's judgment`);\n  }\n\n  const presets = cfg.presets as Record<string, unknown> | undefined;\n  if (!presets || typeof presets !== 'object' || Array.isArray(presets) || Object.keys(presets).length === 0) {\n    throw new SenseError('CONFIG_INVALID', `${configPath}: presets must be a non-empty object of preset name -> { include, exclude?, k?, signals?, where? }`);\n  }\n  for (const [name, value] of Object.entries(presets)) {\n    validatePreset(name, value, configPath);\n  }\n  if (presets.default === undefined) {\n    throw new SenseError('CONFIG_INVALID', `${configPath}: presets must include a \"default\" preset`);\n  }\n\n  if (cfg.queries === undefined) cfg.queries = {};\n  if (typeof cfg.queries !== 'object' || cfg.queries === null || Array.isArray(cfg.queries)) {\n    throw new SenseError('CONFIG_INVALID', `${configPath}: queries must be an object of name -> { sql } or { search }`);\n  }\n  const queries = cfg.queries as Record<string, unknown>;\n  for (const [name, value] of Object.entries(queries)) {\n    validateSavedQuery(name, value, configPath);\n  }\n\n  if (cfg.features !== undefined) {\n    validateFeaturesBlock(cfg.features, configPath);\n  }\n  if (cfg.embed !== undefined) {\n    validateEmbedBlock(cfg.embed, configPath);\n  }\n  if (cfg.content !== undefined) {\n    validateContentBlock(cfg.content, configPath);\n  }\n  if (cfg.store !== undefined) {\n    validateStoreKey(cfg.store, configPath);\n  }\n\n  // vectors' prerequisite is a named model, checked here once embed is known valid (or absent).\n  if (cfg.embed === undefined) {\n    for (const [name, value] of Object.entries(presets)) {\n      const signals = (value as Record<string, unknown>).signals as Record<string, unknown> | undefined;\n      if (signals && 'vectors' in signals) {\n        throw new SenseError('CONFIG_INVALID', `${configPath}: presets.${name}.signals includes \"vectors\", but no \"embed\" block names a model; add one, or drop \"vectors\" from presets.${name}.signals`);\n      }\n    }\n  }\n\n  return cfg as unknown as Config;\n}\n"],"names":["SenseError","SIGNAL_NAMES","SIGNAL_PREREQUISITES","KNOWN_KEYS","Set","KNOWN_PRESET_KEYS","KNOWN_FEATURE_KEYS","KNOWN_EMBED_KEYS","KNOWN_CONTENT_KEYS","SAVED_SEARCH_KEYS","unknownConfigKeys","cfg","Object","keys","filter","k","has","isNonEmptyStringArray","value","Array","isArray","length","every","g","validateLegacyScan","parsed","configPath","scan","include","validateContentBlock","block","unknown","join","tokenize","undefined","trim","validateStoreKey","validateFeaturesBlock","name","v","entries","validateEmbedBlock","embed","model","provider","url","key","languages","some","l","chunkTokens","Number","isInteger","isSignalName","includes","isSignalWeight","isFinite","validatePresetSignals","map","signal","weight","names","requires","validatePreset","preset","exclude","signals","where","validateSavedQuery","JSON","stringify","slice","entry","sql","search","validateConfig","checks","presets","default","queries","features","content","store"],"mappings":"AAAA,SAASA,UAAU,QAAQ,eAAe;AAC1C,SAASC,YAAY,EAAEC,oBAAoB,QAAyB,eAAe;AAGnF,sFAAsF;AACtF,8FAA8F;AAC9F,MAAMC,aAAa,IAAIC,IAAI;IAAC;IAAW;IAAW;IAAW;IAAY;IAAS;IAAW;IAAS;CAAU;AAChH,MAAMC,oBAAoB,IAAID,IAAI;IAAC;IAAW;IAAW;IAAK;IAAW;CAAQ;AACjF,MAAME,qBAAqB,IAAIF,IAAI;IAAC;IAAS;IAAY;IAAQ;CAAO;AACxE,6FAA6F;AAC7F,kFAAkF;AAClF,OAAO,MAAMG,mBAAmB,IAAIH,IAAI;IAAC;IAAS;IAAY;IAAO;IAAO;IAAa;CAAc,EAAE;AACzG,MAAMI,qBAAqB,IAAIJ,IAAI;IAAC;CAAW;AAC/C,MAAMK,oBAAoB,IAAIL,IAAI;IAAC;IAAU;IAAU;IAAW;IAAW;IAAS;CAAI;AAE1F,OAAO,SAASM,kBAAkBC,GAA4B;IAC5D,OAAOC,OAAOC,IAAI,CAACF,KAAKG,MAAM,CAAC,CAACC,IAAM,CAACZ,WAAWa,GAAG,CAACD;AACxD;AAEA,SAASE,sBAAsBC,KAAc;IAC3C,OAAOC,MAAMC,OAAO,CAACF,UAAUA,MAAMG,MAAM,GAAG,KAAKH,MAAMI,KAAK,CAAC,CAACC,IAAM,OAAOA,MAAM,YAAYA,EAAEF,MAAM,GAAG;AAC5G;AAEA,yFAAyF;AACzF,4FAA4F;AAC5F,OAAO,SAASG,mBAAmBC,MAAe,EAAEC,UAAkB;IACpE,MAAMf,MAAM,OAAOc,WAAW,YAAYA,WAAW,QAAQ,CAACN,MAAMC,OAAO,CAACK,UAAWA,SAAqC;IAC5H,MAAME,OAAOhB,gBAAAA,0BAAAA,IAAKgB,IAAI;IACtB,IAAI,CAAChB,OAAO,CAACgB,QAAQ,CAACV,sBAAsBU,KAAKC,OAAO,GAAG;QACzD,MAAM,IAAI5B,WAAW,kBAAkB,GAAG0B,WAAW,wDAAwD,CAAC;IAChH;AACF;AAEA,6FAA6F;AAC7F,yGAAyG;AACzG,SAASG,qBAAqBX,KAAc,EAAEQ,UAAkB;IAC9D,IAAI,OAAOR,UAAU,YAAYA,UAAU,QAAQC,MAAMC,OAAO,CAACF,QAAQ;QACvE,MAAM,IAAIlB,WAAW,kBAAkB,GAAG0B,WAAW,oDAAoD,CAAC;IAC5G;IACA,MAAMI,QAAQZ;IACd,MAAMa,UAAUnB,OAAOC,IAAI,CAACiB,OAAOhB,MAAM,CAAC,CAACC,IAAM,CAACP,mBAAmBQ,GAAG,CAACD;IACzE,IAAIgB,QAAQV,MAAM,GAAG,GAAG;QACtB,MAAM,IAAIrB,WAAW,kBAAkB,GAAG0B,WAAW,6BAA6B,EAAEK,QAAQC,IAAI,CAAC,MAAM,wBAAwB,CAAC;IAClI;IACA,IAAIF,MAAMG,QAAQ,KAAKC,aAAc,CAAA,OAAOJ,MAAMG,QAAQ,KAAK,YAAYH,MAAMG,QAAQ,CAACE,IAAI,OAAO,EAAC,GAAI;QACxG,MAAM,IAAInC,WAAW,kBAAkB,GAAG0B,WAAW,4FAA4F,CAAC;IACpJ;AACF;AAEA,4FAA4F;AAC5F,2FAA2F;AAC3F,gEAAgE;AAChE,SAASU,iBAAiBlB,KAAc,EAAEQ,UAAkB;IAC1D,IAAIR,UAAU,YAAYA,UAAU,UAAU;QAC5C,MAAM,IAAIlB,WAAW,kBAAkB,GAAG0B,WAAW,oCAAoC,CAAC;IAC5F;AACF;AAEA,SAASW,sBAAsBnB,KAAc,EAAEQ,UAAkB;IAC/D,IAAI,OAAOR,UAAU,YAAYA,UAAU,QAAQC,MAAMC,OAAO,CAACF,QAAQ;QACvE,MAAM,IAAIlB,WAAW,kBAAkB,GAAG0B,WAAW,+CAA+C,CAAC;IACvG;IACA,MAAMI,QAAQZ;IACd,MAAMa,UAAUnB,OAAOC,IAAI,CAACiB,OAAOhB,MAAM,CAAC,CAACC,IAAM,CAACT,mBAAmBU,GAAG,CAACD;IACzE,IAAIgB,QAAQV,MAAM,GAAG,GAAG;QACtB,MAAM,IAAIrB,WAAW,kBAAkB,GAAG0B,WAAW,8BAA8B,EAAEK,QAAQC,IAAI,CAAC,MAAM,gEAAgE,CAAC;IAC3K;IACA,KAAK,MAAM,CAACM,MAAMC,EAAE,IAAI3B,OAAO4B,OAAO,CAACV,OAAQ;QAC7C,IAAI,OAAOS,MAAM,WAAW;YAC1B,MAAM,IAAIvC,WAAW,kBAAkB,GAAG0B,WAAW,WAAW,EAAEY,KAAK,kBAAkB,CAAC;QAC5F;IACF;AACF;AAEA,SAASG,mBAAmBvB,KAAc,EAAEQ,UAAkB;IAC5D,IAAI,OAAOR,UAAU,YAAYA,UAAU,QAAQC,MAAMC,OAAO,CAACF,QAAQ;QACvE,MAAM,IAAIlB,WAAW,kBAAkB,GAAG0B,WAAW,0FAA0F,CAAC;IAClJ;IACA,MAAMgB,QAAQxB;IACd,MAAMa,UAAUnB,OAAOC,IAAI,CAAC6B,OAAO5B,MAAM,CAAC,CAACC,IAAM,CAACR,iBAAiBS,GAAG,CAACD;IACvE,IAAIgB,QAAQV,MAAM,GAAG,GAAG;QACtB,MAAM,IAAIrB,WAAW,kBAAkB,GAAG0B,WAAW,2BAA2B,EAAEK,QAAQC,IAAI,CAAC,OAAO;IACxG;IACA,IAAIU,MAAMC,KAAK,KAAKT,aAAa,OAAOQ,MAAMC,KAAK,KAAK,UAAU;QAChE,MAAM,IAAI3C,WAAW,kBAAkB,GAAG0B,WAAW,8BAA8B,CAAC;IACtF;IACA,yFAAyF;IACzF,4EAA4E;IAC5E,IAAIgB,MAAMC,KAAK,KAAKT,aAAaQ,MAAMC,KAAK,KAAK,IAAI;QACnD,MAAM,IAAI3C,WAAW,kBAAkB,GAAG0B,WAAW,oIAAoI,CAAC;IAC5L;IACA,IAAIgB,MAAME,QAAQ,KAAKV,aAAaQ,MAAME,QAAQ,KAAK,YAAYF,MAAME,QAAQ,KAAK,YAAYF,MAAME,QAAQ,KAAK,UAAU;QAC7H,MAAM,IAAI5C,WAAW,kBAAkB,GAAG0B,WAAW,wDAAwD,CAAC;IAChH;IACA,IAAIgB,MAAMG,GAAG,KAAKX,aAAa,OAAOQ,MAAMG,GAAG,KAAK,UAAU;QAC5D,MAAM,IAAI7C,WAAW,kBAAkB,GAAG0B,WAAW,4BAA4B,CAAC;IACpF;IACA,IAAIgB,MAAMI,GAAG,KAAKZ,aAAa,OAAOQ,MAAMI,GAAG,KAAK,UAAU;QAC5D,MAAM,IAAI9C,WAAW,kBAAkB,GAAG0B,WAAW,4BAA4B,CAAC;IACpF;IACA,yFAAyF;IACzF,wDAAwD;IACxD,IAAIgB,MAAME,QAAQ,KAAK,YAAaF,CAAAA,MAAMG,GAAG,KAAKX,aAAaQ,MAAMG,GAAG,KAAK,EAAC,GAAI;QAChF,MAAM,IAAI7C,WAAW,kBAAkB,GAAG0B,WAAW,0FAA0F,CAAC;IAClJ;IACA,IAAIgB,MAAMK,SAAS,KAAKb,aAAc,CAAA,CAACf,MAAMC,OAAO,CAACsB,MAAMK,SAAS,KAAKL,MAAMK,SAAS,CAAC1B,MAAM,KAAK,KAAKqB,MAAMK,SAAS,CAACC,IAAI,CAAC,CAACC,IAAM,OAAOA,MAAM,YAAYA,MAAM,GAAE,GAAI;QACxK,MAAM,IAAIjD,WAAW,kBAAkB,GAAG0B,WAAW,gFAAgF,CAAC;IACxI;IACA,IAAIgB,MAAMQ,WAAW,KAAKhB,aAAc,CAAA,OAAOQ,MAAMQ,WAAW,KAAK,YAAY,CAACC,OAAOC,SAAS,CAACV,MAAMQ,WAAW,KAAKR,MAAMQ,WAAW,IAAI,CAAA,GAAI;QAChJ,MAAM,IAAIlD,WAAW,kBAAkB,GAAG0B,WAAW,8CAA8C,CAAC;IACtG;AACF;AAEA,SAAS2B,aAAanC,KAAc;IAClC,OAAO,OAAOA,UAAU,YAAY,AAACjB,aAAmCqD,QAAQ,CAACpC;AACnF;AAEA,SAASqC,eAAerC,KAAc;IACpC,OAAO,OAAOA,UAAU,YAAYiC,OAAOK,QAAQ,CAACtC,UAAUA,QAAQ;AACxE;AAEA,yFAAyF;AACzF,8FAA8F;AAC9F,iCAAiC;AACjC,SAASuC,sBAAsBnB,IAAY,EAAEpB,KAAc,EAAEQ,UAAkB;IAC7E,IAAI,OAAOR,UAAU,YAAYA,UAAU,QAAQC,MAAMC,OAAO,CAACF,UAAUN,OAAOC,IAAI,CAACK,OAAOG,MAAM,KAAK,GAAG;QAC1G,MAAM,IAAIrB,WACR,kBACA,GAAG0B,WAAW,UAAU,EAAEY,KAAK,2FAA2F,EAAErC,aAAa+B,IAAI,CAAC,MAAM,kJAAkJ,CAAC;IAE3S;IACA,MAAMQ,UAAU5B,OAAO4B,OAAO,CAACtB;IAC/B,MAAMa,UAAUS,QAAQ1B,MAAM,CAAC,CAAC,CAACC,EAAE,GAAK,CAACsC,aAAatC,IAAI2C,GAAG,CAAC,CAAC,CAAC3C,EAAE,GAAKA;IACvE,IAAIgB,QAAQV,MAAM,GAAG,GAAG;QACtB,MAAM,IAAIrB,WAAW,kBAAkB,GAAG0B,WAAW,UAAU,EAAEY,KAAK,iCAAiC,EAAEP,QAAQC,IAAI,CAAC,MAAM,oBAAoB,EAAE/B,aAAa+B,IAAI,CAAC,OAAO;IAC7K;IACA,KAAK,MAAM,CAAC2B,QAAQC,OAAO,IAAIpB,QAAS;QACtC,IAAI,CAACe,eAAeK,SAAS;YAC3B,MAAM,IAAI5D,WAAW,kBAAkB,GAAG0B,WAAW,UAAU,EAAEY,KAAK,SAAS,EAAEqB,OAAO,4BAA4B,CAAC;QACvH;IACF;IACA,MAAME,QAAQ,IAAIzD,IAAIoC,QAAQkB,GAAG,CAAC,CAAC,CAAC3C,EAAE,GAAKA;IAC3C,KAAK,MAAM,CAAC4C,QAAQG,SAAS,IAAIlD,OAAO4B,OAAO,CAACtC,sBAA0D;QACxG,IAAI2D,MAAM7C,GAAG,CAAC2C,WAAW,CAACE,MAAM7C,GAAG,CAAC8C,WAAW;YAC7C,MAAM,IAAI9D,WAAW,kBAAkB,GAAG0B,WAAW,UAAU,EAAEY,KAAK,cAAc,EAAEqB,OAAO,WAAW,EAAEG,SAAS,QAAQ,EAAEA,SAAS,aAAa,EAAExB,KAAK,mBAAmB,EAAEqB,OAAO,CAAC,CAAC;QAC1L;IACF;AACF;AAEA,SAASI,eAAezB,IAAY,EAAEpB,KAAc,EAAEQ,UAAkB;IACtE,IAAI,OAAOR,UAAU,YAAYA,UAAU,QAAQC,MAAMC,OAAO,CAACF,QAAQ;QACvE,MAAM,IAAIlB,WAAW,kBAAkB,GAAG0B,WAAW,UAAU,EAAEY,KAAK,kBAAkB,CAAC;IAC3F;IACA,MAAM0B,SAAS9C;IACf,MAAMa,UAAUnB,OAAOC,IAAI,CAACmD,QAAQlD,MAAM,CAAC,CAACC,IAAM,CAACV,kBAAkBW,GAAG,CAACD;IACzE,IAAIgB,QAAQV,MAAM,GAAG,GAAG;QACtB,MAAM,IAAIrB,WAAW,kBAAkB,GAAG0B,WAAW,UAAU,EAAEY,KAAK,oBAAoB,EAAEP,QAAQC,IAAI,CAAC,MAAM,oDAAoD,CAAC;IACtK;IACA,IAAI,CAACf,sBAAsB+C,OAAOpC,OAAO,GAAG;QAC1C,MAAM,IAAI5B,WAAW,kBAAkB,GAAG0B,WAAW,UAAU,EAAEY,KAAK,kDAAkD,CAAC;IAC3H;IACA,IAAI0B,OAAOC,OAAO,KAAK/B,aAAa,CAACjB,sBAAsB+C,OAAOC,OAAO,GAAG;QAC1E,MAAM,IAAIjE,WAAW,kBAAkB,GAAG0B,WAAW,UAAU,EAAEY,KAAK,kDAAkD,CAAC;IAC3H;IACA,IAAI0B,OAAOE,OAAO,KAAKhC,WAAWuB,sBAAsBnB,MAAM0B,OAAOE,OAAO,EAAExC;IAC9E,IAAIsC,OAAOjD,CAAC,KAAKmB,aAAc,CAAA,OAAO8B,OAAOjD,CAAC,KAAK,YAAY,CAACoC,OAAOC,SAAS,CAACY,OAAOjD,CAAC,KAAKiD,OAAOjD,CAAC,IAAI,CAAA,GAAI;QAC5G,MAAM,IAAIf,WAAW,kBAAkB,GAAG0B,WAAW,UAAU,EAAEY,KAAK,6BAA6B,CAAC;IACtG;IACA,IAAI0B,OAAOG,KAAK,KAAKjC,aAAa,OAAO8B,OAAOG,KAAK,KAAK,UAAU;QAClE,MAAM,IAAInE,WAAW,kBAAkB,GAAG0B,WAAW,UAAU,EAAEY,KAAK,qCAAqC,CAAC;IAC9G;AACF;AAEA,yGAAyG;AACzG,SAAS8B,mBAAmB9B,IAAY,EAAEpB,KAAc,EAAEQ,UAAkB;IAC1E,yFAAyF;IACzF,0EAA0E;IAC1E,IAAI,OAAOR,UAAU,UAAU;QAC7B,MAAM,IAAIlB,WAAW,kBAAkB,GAAG0B,WAAW,UAAU,EAAEY,KAAK,uCAAuC,EAAE+B,KAAKC,SAAS,CAACpD,MAAMG,MAAM,GAAG,KAAK,GAAGH,MAAMqD,KAAK,CAAC,GAAG,IAAI,GAAG,CAAC,GAAGrD,OAAO,+DAA+D,CAAC;IACxP;IACA,IAAI,OAAOA,UAAU,YAAYA,UAAU,QAAQC,MAAMC,OAAO,CAACF,QAAQ;QACvE,MAAM,IAAIlB,WAAW,kBAAkB,GAAG0B,WAAW,UAAU,EAAEY,KAAK,uEAAuE,CAAC;IAChJ;IACA,MAAMkC,QAAQtD;IACd,IAAI,SAASsD,OAAO;QAClB,MAAMzC,UAAUnB,OAAOC,IAAI,CAAC2D,OAAO1D,MAAM,CAAC,CAACC,IAAMA,MAAM;QACvD,IAAIgB,QAAQV,MAAM,GAAG,GAAG;YACtB,MAAM,IAAIrB,WAAW,kBAAkB,GAAG0B,WAAW,UAAU,EAAEY,KAAK,oBAAoB,EAAEP,QAAQC,IAAI,CAAC,MAAM,gCAAgC,CAAC;QAClJ;QACA,IAAI,OAAOwC,MAAMC,GAAG,KAAK,YAAYD,MAAMC,GAAG,CAACtC,IAAI,OAAO,IAAI;YAC5D,MAAM,IAAInC,WAAW,kBAAkB,GAAG0B,WAAW,UAAU,EAAEY,KAAK,+BAA+B,CAAC;QACxG;QACA;IACF;IACA,IAAI,YAAYkC,OAAO;QACrB,MAAMzC,UAAUnB,OAAOC,IAAI,CAAC2D,OAAO1D,MAAM,CAAC,CAACC,IAAM,CAACN,kBAAkBO,GAAG,CAACD;QACxE,IAAIgB,QAAQV,MAAM,GAAG,GAAG;YACtB,MAAM,IAAIrB,WAAW,kBAAkB,GAAG0B,WAAW,UAAU,EAAEY,KAAK,oBAAoB,EAAEP,QAAQC,IAAI,CAAC,MAAM,iEAAiE,CAAC;QACnL;QACA,4EAA4E;QAC5E,IAAI,OAAOwC,MAAME,MAAM,KAAK,YAAYF,MAAME,MAAM,CAACvC,IAAI,OAAO,IAAI;YAClE,MAAM,IAAInC,WAAW,kBAAkB,GAAG0B,WAAW,UAAU,EAAEY,KAAK,8BAA8B,CAAC;QACvG;QACA,IAAIkC,MAAMR,MAAM,KAAK9B,aAAc,CAAA,OAAOsC,MAAMR,MAAM,KAAK,YAAYQ,MAAMR,MAAM,CAAC3C,MAAM,KAAK,CAAA,GAAI;YACjG,MAAM,IAAIrB,WAAW,kBAAkB,GAAG0B,WAAW,UAAU,EAAEY,KAAK,6BAA6B,CAAC;QACtG;QACA,IAAIkC,MAAM5C,OAAO,KAAKM,aAAa,CAACjB,sBAAsBuD,MAAM5C,OAAO,GAAG;YACxE,MAAM,IAAI5B,WAAW,kBAAkB,GAAG0B,WAAW,UAAU,EAAEY,KAAK,kDAAkD,CAAC;QAC3H;QACA,IAAIkC,MAAMP,OAAO,KAAK/B,aAAa,CAACjB,sBAAsBuD,MAAMP,OAAO,GAAG;YACxE,MAAM,IAAIjE,WAAW,kBAAkB,GAAG0B,WAAW,UAAU,EAAEY,KAAK,kDAAkD,CAAC;QAC3H;QACA,IAAIkC,MAAML,KAAK,KAAKjC,aAAa,OAAOsC,MAAML,KAAK,KAAK,UAAU;YAChE,MAAM,IAAInE,WAAW,kBAAkB,GAAG0B,WAAW,UAAU,EAAEY,KAAK,qCAAqC,CAAC;QAC9G;QACA,IAAIkC,MAAMzD,CAAC,KAAKmB,aAAc,CAAA,OAAOsC,MAAMzD,CAAC,KAAK,YAAY,CAACoC,OAAOC,SAAS,CAACoB,MAAMzD,CAAC,KAAKyD,MAAMzD,CAAC,IAAI,CAAA,GAAI;YACxG,MAAM,IAAIf,WAAW,kBAAkB,GAAG0B,WAAW,UAAU,EAAEY,KAAK,6BAA6B,CAAC;QACtG;QACA;IACF;IACA,MAAM,IAAItC,WAAW,kBAAkB,GAAG0B,WAAW,UAAU,EAAEY,KAAK,uEAAuE,CAAC;AAChJ;AAEA,OAAO,SAASqC,eAAelD,MAAe,EAAEC,UAAkB;IAChE,IAAI,OAAOD,WAAW,YAAYA,WAAW,QAAQN,MAAMC,OAAO,CAACK,SAAS;QAC1E,MAAM,IAAIzB,WAAW,kBAAkB,GAAG0B,WAAW,8BAA8B,CAAC;IACtF;IACA,MAAMf,MAAMc;IAEZ,yFAAyF;IACzF,IAAId,IAAIiE,MAAM,KAAK1C,WAAW;QAC5B,MAAM,IAAIlC,WAAW,kBAAkB,GAAG0B,WAAW,yHAAyH,CAAC;IACjL;IAEA,MAAMmD,UAAUlE,IAAIkE,OAAO;IAC3B,IAAI,CAACA,WAAW,OAAOA,YAAY,YAAY1D,MAAMC,OAAO,CAACyD,YAAYjE,OAAOC,IAAI,CAACgE,SAASxD,MAAM,KAAK,GAAG;QAC1G,MAAM,IAAIrB,WAAW,kBAAkB,GAAG0B,WAAW,kGAAkG,CAAC;IAC1J;IACA,KAAK,MAAM,CAACY,MAAMpB,MAAM,IAAIN,OAAO4B,OAAO,CAACqC,SAAU;QACnDd,eAAezB,MAAMpB,OAAOQ;IAC9B;IACA,IAAImD,QAAQC,OAAO,KAAK5C,WAAW;QACjC,MAAM,IAAIlC,WAAW,kBAAkB,GAAG0B,WAAW,yCAAyC,CAAC;IACjG;IAEA,IAAIf,IAAIoE,OAAO,KAAK7C,WAAWvB,IAAIoE,OAAO,GAAG,CAAC;IAC9C,IAAI,OAAOpE,IAAIoE,OAAO,KAAK,YAAYpE,IAAIoE,OAAO,KAAK,QAAQ5D,MAAMC,OAAO,CAACT,IAAIoE,OAAO,GAAG;QACzF,MAAM,IAAI/E,WAAW,kBAAkB,GAAG0B,WAAW,4DAA4D,CAAC;IACpH;IACA,MAAMqD,UAAUpE,IAAIoE,OAAO;IAC3B,KAAK,MAAM,CAACzC,MAAMpB,MAAM,IAAIN,OAAO4B,OAAO,CAACuC,SAAU;QACnDX,mBAAmB9B,MAAMpB,OAAOQ;IAClC;IAEA,IAAIf,IAAIqE,QAAQ,KAAK9C,WAAW;QAC9BG,sBAAsB1B,IAAIqE,QAAQ,EAAEtD;IACtC;IACA,IAAIf,IAAI+B,KAAK,KAAKR,WAAW;QAC3BO,mBAAmB9B,IAAI+B,KAAK,EAAEhB;IAChC;IACA,IAAIf,IAAIsE,OAAO,KAAK/C,WAAW;QAC7BL,qBAAqBlB,IAAIsE,OAAO,EAAEvD;IACpC;IACA,IAAIf,IAAIuE,KAAK,KAAKhD,WAAW;QAC3BE,iBAAiBzB,IAAIuE,KAAK,EAAExD;IAC9B;IAEA,8FAA8F;IAC9F,IAAIf,IAAI+B,KAAK,KAAKR,WAAW;QAC3B,KAAK,MAAM,CAACI,MAAMpB,MAAM,IAAIN,OAAO4B,OAAO,CAACqC,SAAU;YACnD,MAAMX,UAAU,AAAChD,MAAkCgD,OAAO;YAC1D,IAAIA,WAAW,aAAaA,SAAS;gBACnC,MAAM,IAAIlE,WAAW,kBAAkB,GAAG0B,WAAW,UAAU,EAAEY,KAAK,yGAAyG,EAAEA,KAAK,QAAQ,CAAC;YACjM;QACF;IACF;IAEA,OAAO3B;AACT"}
+{"version":3,"sources":["/Users/kevin/Dev/OpenSource/ai/sensemaking/src/config/validate.ts"],"sourcesContent":["import { SenseError } from '../errors.ts';\nimport { SIGNAL_NAMES, SIGNAL_PREREQUISITES, type SignalName } from './signals.ts';\nimport { type Config, STORE_NAMES, type StoreName } from './types.ts';\n\n// Shape check for hand-edited files, so a typo names itself instead of surfacing as a\n// TypeError. Unknown top-level keys warn (forward compat); unknown keys inside a block error.\nconst KNOWN_KEYS = new Set(['$schema', 'version', 'presets', 'features', 'embed', 'content', 'store', 'queries']);\nconst KNOWN_PRESET_KEYS = new Set(['include', 'exclude', 'k', 'signals', 'where']);\nconst KNOWN_FEATURE_KEYS = new Set(['links', 'sections', 'tags', 'rank']);\n// Exported so a docs test can assert every key here has a schema.json property -- one source\n// of truth for what the embed block accepts, checked against the other for drift.\nexport const KNOWN_EMBED_KEYS = new Set(['model', 'provider', 'url', 'key', 'languages', 'chunkTokens']);\nconst KNOWN_CONTENT_KEYS = new Set(['tokenize']);\nconst SAVED_SEARCH_KEYS = new Set(['search', 'preset', 'include', 'exclude', 'where', 'k']);\n\nexport function unknownConfigKeys(cfg: Record<string, unknown>): string[] {\n  return Object.keys(cfg).filter((k) => !KNOWN_KEYS.has(k));\n}\n\nfunction isNonEmptyStringArray(value: unknown): value is string[] {\n  return Array.isArray(value) && value.length > 0 && value.every((g) => typeof g === 'string' && g.length > 0);\n}\n\n// Pre-v3 shape check, just enough for migrateConfig to run safely (it reads scan.include\n// directly). The full v3 shape is checked by validateConfig once migration has produced it.\nexport function validateLegacyScan(parsed: unknown, configPath: string): void {\n  const cfg = typeof parsed === 'object' && parsed !== null && !Array.isArray(parsed) ? (parsed as Record<string, unknown>) : null;\n  const scan = cfg?.scan as { include?: unknown } | undefined;\n  if (!cfg || !scan || !isNonEmptyStringArray(scan.include)) {\n    throw new SenseError('CONFIG_INVALID', `${configPath}: scan.include must be a non-empty array of glob strings`);\n  }\n}\n\n// Only the shape is checked here. Whether the linked SQLite accepts the tokenizer is settled\n// by probing it in store/sqlite/open.ts, so this never has to carry a table of which version added what.\nfunction validateContentBlock(value: unknown, configPath: string): void {\n  if (typeof value !== 'object' || value === null || Array.isArray(value)) {\n    throw new SenseError('CONFIG_INVALID', `${configPath}: content must be an object of { tokenize?: string }`);\n  }\n  const block = value as Record<string, unknown>;\n  const unknown = Object.keys(block).filter((k) => !KNOWN_CONTENT_KEYS.has(k));\n  if (unknown.length > 0) {\n    throw new SenseError('CONFIG_INVALID', `${configPath}: content has unknown key(s) ${unknown.join(', ')}; content takes tokenize`);\n  }\n  if (block.tokenize !== undefined && (typeof block.tokenize !== 'string' || block.tokenize.trim() === '')) {\n    throw new SenseError('CONFIG_INVALID', `${configPath}: content.tokenize must be a non-empty string, e.g. \"trigram\" or \"unicode61 tokenchars '-_'\"`);\n  }\n}\n\nfunction validateStoreKey(value: unknown, configPath: string): void {\n  if (!STORE_NAMES.includes(value as StoreName)) {\n    throw new SenseError('CONFIG_INVALID', `${configPath}: store must be ${STORE_NAMES.map((name) => `\"${name}\"`).join(' or ')}`);\n  }\n}\n\nfunction validateFeaturesBlock(value: unknown, configPath: string): void {\n  if (typeof value !== 'object' || value === null || Array.isArray(value)) {\n    throw new SenseError('CONFIG_INVALID', `${configPath}: features must be an object of name -> boolean`);\n  }\n  const block = value as Record<string, unknown>;\n  const unknown = Object.keys(block).filter((k) => !KNOWN_FEATURE_KEYS.has(k));\n  if (unknown.length > 0) {\n    throw new SenseError('CONFIG_INVALID', `${configPath}: features has unknown key(s) ${unknown.join(', ')}; embed is not a features key -- see the top-level \"embed\" block`);\n  }\n  for (const [name, v] of Object.entries(block)) {\n    if (typeof v !== 'boolean') {\n      throw new SenseError('CONFIG_INVALID', `${configPath}: features.${name} must be a boolean`);\n    }\n  }\n}\n\nfunction validateEmbedBlock(value: unknown, configPath: string): void {\n  if (typeof value !== 'object' || value === null || Array.isArray(value)) {\n    throw new SenseError('CONFIG_INVALID', `${configPath}: embed must be an object of { model?, provider?: \"static\"|\"openai\"|\"cohere\", url?, key? }`);\n  }\n  const embed = value as Record<string, unknown>;\n  const unknown = Object.keys(embed).filter((k) => !KNOWN_EMBED_KEYS.has(k));\n  if (unknown.length > 0) {\n    throw new SenseError('CONFIG_INVALID', `${configPath}: embed has unknown key(s) ${unknown.join(', ')}`);\n  }\n  if (embed.model !== undefined && typeof embed.model !== 'string') {\n    throw new SenseError('CONFIG_INVALID', `${configPath}: embed.model must be a string`);\n  }\n  // The block exists to name a model; provider settings without one would read as \"vectors\n  // configured\" while embedding stays off, which is the ambiguity v4 removed.\n  if (embed.model === undefined || embed.model === '') {\n    throw new SenseError('CONFIG_INVALID', `${configPath}: embed.model is required when the \"embed\" block is present (it is what turns vectors on); remove the block to index without vectors`);\n  }\n  if (embed.provider !== undefined && embed.provider !== 'static' && embed.provider !== 'openai' && embed.provider !== 'cohere') {\n    throw new SenseError('CONFIG_INVALID', `${configPath}: embed.provider must be \"static\", \"openai\", or \"cohere\"`);\n  }\n  if (embed.url !== undefined && typeof embed.url !== 'string') {\n    throw new SenseError('CONFIG_INVALID', `${configPath}: embed.url must be a string`);\n  }\n  if (embed.key !== undefined && typeof embed.key !== 'string') {\n    throw new SenseError('CONFIG_INVALID', `${configPath}: embed.key must be a string`);\n  }\n  // openai has no default host, so a missing url would otherwise only surface at the first\n  // vector search; a config error belongs at config time.\n  if (embed.provider === 'openai' && (embed.url === undefined || embed.url === '')) {\n    throw new SenseError('CONFIG_INVALID', `${configPath}: embed.provider \"openai\" requires embed.url (e.g. \"http://localhost:11434/v1\" for Ollama)`);\n  }\n  if (embed.languages !== undefined && (!Array.isArray(embed.languages) || embed.languages.length === 0 || embed.languages.some((l) => typeof l !== 'string' || l === ''))) {\n    throw new SenseError('CONFIG_INVALID', `${configPath}: embed.languages must be a non-empty array of language tags (e.g. [\"en\", \"zh\"])`);\n  }\n  if (embed.chunkTokens !== undefined && (typeof embed.chunkTokens !== 'number' || !Number.isInteger(embed.chunkTokens) || embed.chunkTokens <= 0)) {\n    throw new SenseError('CONFIG_INVALID', `${configPath}: embed.chunkTokens must be a positive integer`);\n  }\n}\n\nfunction isSignalName(value: unknown): value is SignalName {\n  return typeof value === 'string' && (SIGNAL_NAMES as readonly string[]).includes(value);\n}\n\nfunction isSignalWeight(value: unknown): value is number {\n  return typeof value === 'number' && Number.isFinite(value) && value > 0;\n}\n\n// Shape (an object of signal name -> weight) plus the one cross-signal prerequisite that doesn't\n// need the top-level embed block: links is seeded by word-match rows, so declaring it without words is always empty.\nfunction validatePresetSignals(name: string, value: unknown, configPath: string): void {\n  if (typeof value !== 'object' || value === null || Array.isArray(value) || Object.keys(value).length === 0) {\n    throw new SenseError(\n      'CONFIG_INVALID',\n      `${configPath}: presets.${name}.signals must be a non-empty object of signal name -> weight (a finite number > 0), naming ${SIGNAL_NAMES.join(', ')} -- a preset that searches nothing is a mistake; name at least \"words\", or omit \"signals\" to use every signal whose prerequisites hold at weight 1`\n    );\n  }\n  const entries = Object.entries(value as Record<string, unknown>);\n  const unknown = entries.filter(([k]) => !isSignalName(k)).map(([k]) => k);\n  if (unknown.length > 0) {\n    throw new SenseError('CONFIG_INVALID', `${configPath}: presets.${name}.signals names unknown signal(s) ${unknown.join(', ')}; valid signals are ${SIGNAL_NAMES.join(', ')}`);\n  }\n  for (const [signal, weight] of entries) {\n    if (!isSignalWeight(weight)) {\n      throw new SenseError('CONFIG_INVALID', `${configPath}: presets.${name}.signals.${signal} must be a finite number > 0`);\n    }\n  }\n  const names = new Set(entries.map(([k]) => k));\n  for (const [signal, requires] of Object.entries(SIGNAL_PREREQUISITES) as Array<[SignalName, SignalName]>) {\n    if (names.has(signal) && !names.has(requires)) {\n      throw new SenseError('CONFIG_INVALID', `${configPath}: presets.${name}.signals has \"${signal}\" without \"${requires}\"; add \"${requires}\" to presets.${name}.signals, or drop \"${signal}\"`);\n    }\n  }\n}\n\nfunction validatePreset(name: string, value: unknown, configPath: string): void {\n  if (typeof value !== 'object' || value === null || Array.isArray(value)) {\n    throw new SenseError('CONFIG_INVALID', `${configPath}: presets.${name} must be an object`);\n  }\n  const preset = value as Record<string, unknown>;\n  const unknown = Object.keys(preset).filter((k) => !KNOWN_PRESET_KEYS.has(k));\n  if (unknown.length > 0) {\n    throw new SenseError('CONFIG_INVALID', `${configPath}: presets.${name} has unknown key(s) ${unknown.join(', ')}; a preset takes include, exclude, k, signals, where`);\n  }\n  if (!isNonEmptyStringArray(preset.include)) {\n    throw new SenseError('CONFIG_INVALID', `${configPath}: presets.${name}.include must be a non-empty array of glob strings`);\n  }\n  if (preset.exclude !== undefined && !isNonEmptyStringArray(preset.exclude)) {\n    throw new SenseError('CONFIG_INVALID', `${configPath}: presets.${name}.exclude must be a non-empty array of glob strings`);\n  }\n  if (preset.signals !== undefined) validatePresetSignals(name, preset.signals, configPath);\n  if (preset.k !== undefined && (typeof preset.k !== 'number' || !Number.isInteger(preset.k) || preset.k <= 0)) {\n    throw new SenseError('CONFIG_INVALID', `${configPath}: presets.${name}.k must be a positive integer`);\n  }\n  if (preset.where !== undefined && typeof preset.where !== 'string') {\n    throw new SenseError('CONFIG_INVALID', `${configPath}: presets.${name}.where must be a SQL condition string`);\n  }\n}\n\n// A queries.<name> entry: { sql } or a saved search { search, preset?, include?, exclude?, where?, k? }.\nfunction validateSavedQuery(name: string, value: unknown, configPath: string): void {\n  // A bare string is rejected rather than inferred to be SQL: an entry says which of the two\n  // verbs it runs, the same choice the CLI makes explicit.\n  if (typeof value === 'string') {\n    throw new SenseError('CONFIG_INVALID', `${configPath}: queries.${name} must say which verb it runs: { \"sql\": ${JSON.stringify(value.length > 40 ? `${value.slice(0, 37)}...` : value)} } to run it as SQL, or { \"search\": \"...\" } for a ranked search`);\n  }\n  if (typeof value !== 'object' || value === null || Array.isArray(value)) {\n    throw new SenseError('CONFIG_INVALID', `${configPath}: queries.${name} must be { sql } or { search, preset?, include?, exclude?, where?, k? }`);\n  }\n  const entry = value as Record<string, unknown>;\n  if ('sql' in entry) {\n    const unknown = Object.keys(entry).filter((k) => k !== 'sql');\n    if (unknown.length > 0) {\n      throw new SenseError('CONFIG_INVALID', `${configPath}: queries.${name} has unknown key(s) ${unknown.join(', ')}; a { sql } query takes only sql`);\n    }\n    if (typeof entry.sql !== 'string' || entry.sql.trim() === '') {\n      throw new SenseError('CONFIG_INVALID', `${configPath}: queries.${name}.sql must be a non-empty string`);\n    }\n    return;\n  }\n  if ('search' in entry) {\n    const unknown = Object.keys(entry).filter((k) => !SAVED_SEARCH_KEYS.has(k));\n    if (unknown.length > 0) {\n      throw new SenseError('CONFIG_INVALID', `${configPath}: queries.${name} has unknown key(s) ${unknown.join(', ')}; a saved search takes search, preset, include, exclude, where, k`);\n    }\n    // A saved entry saves a question; a scope without a question is just flags.\n    if (typeof entry.search !== 'string' || entry.search.trim() === '') {\n      throw new SenseError('CONFIG_INVALID', `${configPath}: queries.${name}.search must be non-empty text`);\n    }\n    if (entry.preset !== undefined && (typeof entry.preset !== 'string' || entry.preset.length === 0)) {\n      throw new SenseError('CONFIG_INVALID', `${configPath}: queries.${name}.preset must be a preset name`);\n    }\n    if (entry.include !== undefined && !isNonEmptyStringArray(entry.include)) {\n      throw new SenseError('CONFIG_INVALID', `${configPath}: queries.${name}.include must be a non-empty array of glob strings`);\n    }\n    if (entry.exclude !== undefined && !isNonEmptyStringArray(entry.exclude)) {\n      throw new SenseError('CONFIG_INVALID', `${configPath}: queries.${name}.exclude must be a non-empty array of glob strings`);\n    }\n    if (entry.where !== undefined && typeof entry.where !== 'string') {\n      throw new SenseError('CONFIG_INVALID', `${configPath}: queries.${name}.where must be a SQL condition string`);\n    }\n    if (entry.k !== undefined && (typeof entry.k !== 'number' || !Number.isInteger(entry.k) || entry.k <= 0)) {\n      throw new SenseError('CONFIG_INVALID', `${configPath}: queries.${name}.k must be a positive integer`);\n    }\n    return;\n  }\n  throw new SenseError('CONFIG_INVALID', `${configPath}: queries.${name} must be { sql } or { search, preset?, include?, exclude?, where?, k? }`);\n}\n\nexport function validateConfig(parsed: unknown, configPath: string): Config {\n  if (typeof parsed !== 'object' || parsed === null || Array.isArray(parsed)) {\n    throw new SenseError('CONFIG_INVALID', `${configPath}: config must be a JSON object`);\n  }\n  const cfg = parsed as Record<string, unknown>;\n\n  // `checks` is rejected by name, not warned: silence would hide that assertions are gone.\n  if (cfg.checks !== undefined) {\n    throw new SenseError('CONFIG_INVALID', `${configPath}: checks was removed in v3 -- sense check no longer asserts on saved queries; a returned row set is the reader's judgment`);\n  }\n\n  const presets = cfg.presets as Record<string, unknown> | undefined;\n  if (!presets || typeof presets !== 'object' || Array.isArray(presets) || Object.keys(presets).length === 0) {\n    throw new SenseError('CONFIG_INVALID', `${configPath}: presets must be a non-empty object of preset name -> { include, exclude?, k?, signals?, where? }`);\n  }\n  for (const [name, value] of Object.entries(presets)) {\n    validatePreset(name, value, configPath);\n  }\n  if (presets.default === undefined) {\n    throw new SenseError('CONFIG_INVALID', `${configPath}: presets must include a \"default\" preset`);\n  }\n\n  if (cfg.queries === undefined) cfg.queries = {};\n  if (typeof cfg.queries !== 'object' || cfg.queries === null || Array.isArray(cfg.queries)) {\n    throw new SenseError('CONFIG_INVALID', `${configPath}: queries must be an object of name -> { sql } or { search }`);\n  }\n  const queries = cfg.queries as Record<string, unknown>;\n  for (const [name, value] of Object.entries(queries)) {\n    validateSavedQuery(name, value, configPath);\n  }\n\n  if (cfg.features !== undefined) {\n    validateFeaturesBlock(cfg.features, configPath);\n  }\n  if (cfg.embed !== undefined) {\n    validateEmbedBlock(cfg.embed, configPath);\n  }\n  if (cfg.content !== undefined) {\n    validateContentBlock(cfg.content, configPath);\n  }\n  if (cfg.store !== undefined) {\n    validateStoreKey(cfg.store, configPath);\n  }\n\n  // vectors' prerequisite is a named model, checked here once embed is known valid (or absent).\n  if (cfg.embed === undefined) {\n    for (const [name, value] of Object.entries(presets)) {\n      const signals = (value as Record<string, unknown>).signals as Record<string, unknown> | undefined;\n      if (signals && 'vectors' in signals) {\n        throw new SenseError('CONFIG_INVALID', `${configPath}: presets.${name}.signals includes \"vectors\", but no \"embed\" block names a model; add one, or drop \"vectors\" from presets.${name}.signals`);\n      }\n    }\n  }\n\n  return cfg as unknown as Config;\n}\n"],"names":["SenseError","SIGNAL_NAMES","SIGNAL_PREREQUISITES","STORE_NAMES","KNOWN_KEYS","Set","KNOWN_PRESET_KEYS","KNOWN_FEATURE_KEYS","KNOWN_EMBED_KEYS","KNOWN_CONTENT_KEYS","SAVED_SEARCH_KEYS","unknownConfigKeys","cfg","Object","keys","filter","k","has","isNonEmptyStringArray","value","Array","isArray","length","every","g","validateLegacyScan","parsed","configPath","scan","include","validateContentBlock","block","unknown","join","tokenize","undefined","trim","validateStoreKey","includes","map","name","validateFeaturesBlock","v","entries","validateEmbedBlock","embed","model","provider","url","key","languages","some","l","chunkTokens","Number","isInteger","isSignalName","isSignalWeight","isFinite","validatePresetSignals","signal","weight","names","requires","validatePreset","preset","exclude","signals","where","validateSavedQuery","JSON","stringify","slice","entry","sql","search","validateConfig","checks","presets","default","queries","features","content","store"],"mappings":"AAAA,SAASA,UAAU,QAAQ,eAAe;AAC1C,SAASC,YAAY,EAAEC,oBAAoB,QAAyB,eAAe;AACnF,SAAsBC,WAAW,QAAwB,aAAa;AAEtE,sFAAsF;AACtF,8FAA8F;AAC9F,MAAMC,aAAa,IAAIC,IAAI;IAAC;IAAW;IAAW;IAAW;IAAY;IAAS;IAAW;IAAS;CAAU;AAChH,MAAMC,oBAAoB,IAAID,IAAI;IAAC;IAAW;IAAW;IAAK;IAAW;CAAQ;AACjF,MAAME,qBAAqB,IAAIF,IAAI;IAAC;IAAS;IAAY;IAAQ;CAAO;AACxE,6FAA6F;AAC7F,kFAAkF;AAClF,OAAO,MAAMG,mBAAmB,IAAIH,IAAI;IAAC;IAAS;IAAY;IAAO;IAAO;IAAa;CAAc,EAAE;AACzG,MAAMI,qBAAqB,IAAIJ,IAAI;IAAC;CAAW;AAC/C,MAAMK,oBAAoB,IAAIL,IAAI;IAAC;IAAU;IAAU;IAAW;IAAW;IAAS;CAAI;AAE1F,OAAO,SAASM,kBAAkBC,GAA4B;IAC5D,OAAOC,OAAOC,IAAI,CAACF,KAAKG,MAAM,CAAC,CAACC,IAAM,CAACZ,WAAWa,GAAG,CAACD;AACxD;AAEA,SAASE,sBAAsBC,KAAc;IAC3C,OAAOC,MAAMC,OAAO,CAACF,UAAUA,MAAMG,MAAM,GAAG,KAAKH,MAAMI,KAAK,CAAC,CAACC,IAAM,OAAOA,MAAM,YAAYA,EAAEF,MAAM,GAAG;AAC5G;AAEA,yFAAyF;AACzF,4FAA4F;AAC5F,OAAO,SAASG,mBAAmBC,MAAe,EAAEC,UAAkB;IACpE,MAAMf,MAAM,OAAOc,WAAW,YAAYA,WAAW,QAAQ,CAACN,MAAMC,OAAO,CAACK,UAAWA,SAAqC;IAC5H,MAAME,OAAOhB,gBAAAA,0BAAAA,IAAKgB,IAAI;IACtB,IAAI,CAAChB,OAAO,CAACgB,QAAQ,CAACV,sBAAsBU,KAAKC,OAAO,GAAG;QACzD,MAAM,IAAI7B,WAAW,kBAAkB,GAAG2B,WAAW,wDAAwD,CAAC;IAChH;AACF;AAEA,6FAA6F;AAC7F,yGAAyG;AACzG,SAASG,qBAAqBX,KAAc,EAAEQ,UAAkB;IAC9D,IAAI,OAAOR,UAAU,YAAYA,UAAU,QAAQC,MAAMC,OAAO,CAACF,QAAQ;QACvE,MAAM,IAAInB,WAAW,kBAAkB,GAAG2B,WAAW,oDAAoD,CAAC;IAC5G;IACA,MAAMI,QAAQZ;IACd,MAAMa,UAAUnB,OAAOC,IAAI,CAACiB,OAAOhB,MAAM,CAAC,CAACC,IAAM,CAACP,mBAAmBQ,GAAG,CAACD;IACzE,IAAIgB,QAAQV,MAAM,GAAG,GAAG;QACtB,MAAM,IAAItB,WAAW,kBAAkB,GAAG2B,WAAW,6BAA6B,EAAEK,QAAQC,IAAI,CAAC,MAAM,wBAAwB,CAAC;IAClI;IACA,IAAIF,MAAMG,QAAQ,KAAKC,aAAc,CAAA,OAAOJ,MAAMG,QAAQ,KAAK,YAAYH,MAAMG,QAAQ,CAACE,IAAI,OAAO,EAAC,GAAI;QACxG,MAAM,IAAIpC,WAAW,kBAAkB,GAAG2B,WAAW,4FAA4F,CAAC;IACpJ;AACF;AAEA,SAASU,iBAAiBlB,KAAc,EAAEQ,UAAkB;IAC1D,IAAI,CAACxB,YAAYmC,QAAQ,CAACnB,QAAqB;QAC7C,MAAM,IAAInB,WAAW,kBAAkB,GAAG2B,WAAW,gBAAgB,EAAExB,YAAYoC,GAAG,CAAC,CAACC,OAAS,CAAC,CAAC,EAAEA,KAAK,CAAC,CAAC,EAAEP,IAAI,CAAC,SAAS;IAC9H;AACF;AAEA,SAASQ,sBAAsBtB,KAAc,EAAEQ,UAAkB;IAC/D,IAAI,OAAOR,UAAU,YAAYA,UAAU,QAAQC,MAAMC,OAAO,CAACF,QAAQ;QACvE,MAAM,IAAInB,WAAW,kBAAkB,GAAG2B,WAAW,+CAA+C,CAAC;IACvG;IACA,MAAMI,QAAQZ;IACd,MAAMa,UAAUnB,OAAOC,IAAI,CAACiB,OAAOhB,MAAM,CAAC,CAACC,IAAM,CAACT,mBAAmBU,GAAG,CAACD;IACzE,IAAIgB,QAAQV,MAAM,GAAG,GAAG;QACtB,MAAM,IAAItB,WAAW,kBAAkB,GAAG2B,WAAW,8BAA8B,EAAEK,QAAQC,IAAI,CAAC,MAAM,gEAAgE,CAAC;IAC3K;IACA,KAAK,MAAM,CAACO,MAAME,EAAE,IAAI7B,OAAO8B,OAAO,CAACZ,OAAQ;QAC7C,IAAI,OAAOW,MAAM,WAAW;YAC1B,MAAM,IAAI1C,WAAW,kBAAkB,GAAG2B,WAAW,WAAW,EAAEa,KAAK,kBAAkB,CAAC;QAC5F;IACF;AACF;AAEA,SAASI,mBAAmBzB,KAAc,EAAEQ,UAAkB;IAC5D,IAAI,OAAOR,UAAU,YAAYA,UAAU,QAAQC,MAAMC,OAAO,CAACF,QAAQ;QACvE,MAAM,IAAInB,WAAW,kBAAkB,GAAG2B,WAAW,0FAA0F,CAAC;IAClJ;IACA,MAAMkB,QAAQ1B;IACd,MAAMa,UAAUnB,OAAOC,IAAI,CAAC+B,OAAO9B,MAAM,CAAC,CAACC,IAAM,CAACR,iBAAiBS,GAAG,CAACD;IACvE,IAAIgB,QAAQV,MAAM,GAAG,GAAG;QACtB,MAAM,IAAItB,WAAW,kBAAkB,GAAG2B,WAAW,2BAA2B,EAAEK,QAAQC,IAAI,CAAC,OAAO;IACxG;IACA,IAAIY,MAAMC,KAAK,KAAKX,aAAa,OAAOU,MAAMC,KAAK,KAAK,UAAU;QAChE,MAAM,IAAI9C,WAAW,kBAAkB,GAAG2B,WAAW,8BAA8B,CAAC;IACtF;IACA,yFAAyF;IACzF,4EAA4E;IAC5E,IAAIkB,MAAMC,KAAK,KAAKX,aAAaU,MAAMC,KAAK,KAAK,IAAI;QACnD,MAAM,IAAI9C,WAAW,kBAAkB,GAAG2B,WAAW,oIAAoI,CAAC;IAC5L;IACA,IAAIkB,MAAME,QAAQ,KAAKZ,aAAaU,MAAME,QAAQ,KAAK,YAAYF,MAAME,QAAQ,KAAK,YAAYF,MAAME,QAAQ,KAAK,UAAU;QAC7H,MAAM,IAAI/C,WAAW,kBAAkB,GAAG2B,WAAW,wDAAwD,CAAC;IAChH;IACA,IAAIkB,MAAMG,GAAG,KAAKb,aAAa,OAAOU,MAAMG,GAAG,KAAK,UAAU;QAC5D,MAAM,IAAIhD,WAAW,kBAAkB,GAAG2B,WAAW,4BAA4B,CAAC;IACpF;IACA,IAAIkB,MAAMI,GAAG,KAAKd,aAAa,OAAOU,MAAMI,GAAG,KAAK,UAAU;QAC5D,MAAM,IAAIjD,WAAW,kBAAkB,GAAG2B,WAAW,4BAA4B,CAAC;IACpF;IACA,yFAAyF;IACzF,wDAAwD;IACxD,IAAIkB,MAAME,QAAQ,KAAK,YAAaF,CAAAA,MAAMG,GAAG,KAAKb,aAAaU,MAAMG,GAAG,KAAK,EAAC,GAAI;QAChF,MAAM,IAAIhD,WAAW,kBAAkB,GAAG2B,WAAW,0FAA0F,CAAC;IAClJ;IACA,IAAIkB,MAAMK,SAAS,KAAKf,aAAc,CAAA,CAACf,MAAMC,OAAO,CAACwB,MAAMK,SAAS,KAAKL,MAAMK,SAAS,CAAC5B,MAAM,KAAK,KAAKuB,MAAMK,SAAS,CAACC,IAAI,CAAC,CAACC,IAAM,OAAOA,MAAM,YAAYA,MAAM,GAAE,GAAI;QACxK,MAAM,IAAIpD,WAAW,kBAAkB,GAAG2B,WAAW,gFAAgF,CAAC;IACxI;IACA,IAAIkB,MAAMQ,WAAW,KAAKlB,aAAc,CAAA,OAAOU,MAAMQ,WAAW,KAAK,YAAY,CAACC,OAAOC,SAAS,CAACV,MAAMQ,WAAW,KAAKR,MAAMQ,WAAW,IAAI,CAAA,GAAI;QAChJ,MAAM,IAAIrD,WAAW,kBAAkB,GAAG2B,WAAW,8CAA8C,CAAC;IACtG;AACF;AAEA,SAAS6B,aAAarC,KAAc;IAClC,OAAO,OAAOA,UAAU,YAAY,AAAClB,aAAmCqC,QAAQ,CAACnB;AACnF;AAEA,SAASsC,eAAetC,KAAc;IACpC,OAAO,OAAOA,UAAU,YAAYmC,OAAOI,QAAQ,CAACvC,UAAUA,QAAQ;AACxE;AAEA,iGAAiG;AACjG,qHAAqH;AACrH,SAASwC,sBAAsBnB,IAAY,EAAErB,KAAc,EAAEQ,UAAkB;IAC7E,IAAI,OAAOR,UAAU,YAAYA,UAAU,QAAQC,MAAMC,OAAO,CAACF,UAAUN,OAAOC,IAAI,CAACK,OAAOG,MAAM,KAAK,GAAG;QAC1G,MAAM,IAAItB,WACR,kBACA,GAAG2B,WAAW,UAAU,EAAEa,KAAK,2FAA2F,EAAEvC,aAAagC,IAAI,CAAC,MAAM,kJAAkJ,CAAC;IAE3S;IACA,MAAMU,UAAU9B,OAAO8B,OAAO,CAACxB;IAC/B,MAAMa,UAAUW,QAAQ5B,MAAM,CAAC,CAAC,CAACC,EAAE,GAAK,CAACwC,aAAaxC,IAAIuB,GAAG,CAAC,CAAC,CAACvB,EAAE,GAAKA;IACvE,IAAIgB,QAAQV,MAAM,GAAG,GAAG;QACtB,MAAM,IAAItB,WAAW,kBAAkB,GAAG2B,WAAW,UAAU,EAAEa,KAAK,iCAAiC,EAAER,QAAQC,IAAI,CAAC,MAAM,oBAAoB,EAAEhC,aAAagC,IAAI,CAAC,OAAO;IAC7K;IACA,KAAK,MAAM,CAAC2B,QAAQC,OAAO,IAAIlB,QAAS;QACtC,IAAI,CAACc,eAAeI,SAAS;YAC3B,MAAM,IAAI7D,WAAW,kBAAkB,GAAG2B,WAAW,UAAU,EAAEa,KAAK,SAAS,EAAEoB,OAAO,4BAA4B,CAAC;QACvH;IACF;IACA,MAAME,QAAQ,IAAIzD,IAAIsC,QAAQJ,GAAG,CAAC,CAAC,CAACvB,EAAE,GAAKA;IAC3C,KAAK,MAAM,CAAC4C,QAAQG,SAAS,IAAIlD,OAAO8B,OAAO,CAACzC,sBAA0D;QACxG,IAAI4D,MAAM7C,GAAG,CAAC2C,WAAW,CAACE,MAAM7C,GAAG,CAAC8C,WAAW;YAC7C,MAAM,IAAI/D,WAAW,kBAAkB,GAAG2B,WAAW,UAAU,EAAEa,KAAK,cAAc,EAAEoB,OAAO,WAAW,EAAEG,SAAS,QAAQ,EAAEA,SAAS,aAAa,EAAEvB,KAAK,mBAAmB,EAAEoB,OAAO,CAAC,CAAC;QAC1L;IACF;AACF;AAEA,SAASI,eAAexB,IAAY,EAAErB,KAAc,EAAEQ,UAAkB;IACtE,IAAI,OAAOR,UAAU,YAAYA,UAAU,QAAQC,MAAMC,OAAO,CAACF,QAAQ;QACvE,MAAM,IAAInB,WAAW,kBAAkB,GAAG2B,WAAW,UAAU,EAAEa,KAAK,kBAAkB,CAAC;IAC3F;IACA,MAAMyB,SAAS9C;IACf,MAAMa,UAAUnB,OAAOC,IAAI,CAACmD,QAAQlD,MAAM,CAAC,CAACC,IAAM,CAACV,kBAAkBW,GAAG,CAACD;IACzE,IAAIgB,QAAQV,MAAM,GAAG,GAAG;QACtB,MAAM,IAAItB,WAAW,kBAAkB,GAAG2B,WAAW,UAAU,EAAEa,KAAK,oBAAoB,EAAER,QAAQC,IAAI,CAAC,MAAM,oDAAoD,CAAC;IACtK;IACA,IAAI,CAACf,sBAAsB+C,OAAOpC,OAAO,GAAG;QAC1C,MAAM,IAAI7B,WAAW,kBAAkB,GAAG2B,WAAW,UAAU,EAAEa,KAAK,kDAAkD,CAAC;IAC3H;IACA,IAAIyB,OAAOC,OAAO,KAAK/B,aAAa,CAACjB,sBAAsB+C,OAAOC,OAAO,GAAG;QAC1E,MAAM,IAAIlE,WAAW,kBAAkB,GAAG2B,WAAW,UAAU,EAAEa,KAAK,kDAAkD,CAAC;IAC3H;IACA,IAAIyB,OAAOE,OAAO,KAAKhC,WAAWwB,sBAAsBnB,MAAMyB,OAAOE,OAAO,EAAExC;IAC9E,IAAIsC,OAAOjD,CAAC,KAAKmB,aAAc,CAAA,OAAO8B,OAAOjD,CAAC,KAAK,YAAY,CAACsC,OAAOC,SAAS,CAACU,OAAOjD,CAAC,KAAKiD,OAAOjD,CAAC,IAAI,CAAA,GAAI;QAC5G,MAAM,IAAIhB,WAAW,kBAAkB,GAAG2B,WAAW,UAAU,EAAEa,KAAK,6BAA6B,CAAC;IACtG;IACA,IAAIyB,OAAOG,KAAK,KAAKjC,aAAa,OAAO8B,OAAOG,KAAK,KAAK,UAAU;QAClE,MAAM,IAAIpE,WAAW,kBAAkB,GAAG2B,WAAW,UAAU,EAAEa,KAAK,qCAAqC,CAAC;IAC9G;AACF;AAEA,yGAAyG;AACzG,SAAS6B,mBAAmB7B,IAAY,EAAErB,KAAc,EAAEQ,UAAkB;IAC1E,2FAA2F;IAC3F,yDAAyD;IACzD,IAAI,OAAOR,UAAU,UAAU;QAC7B,MAAM,IAAInB,WAAW,kBAAkB,GAAG2B,WAAW,UAAU,EAAEa,KAAK,uCAAuC,EAAE8B,KAAKC,SAAS,CAACpD,MAAMG,MAAM,GAAG,KAAK,GAAGH,MAAMqD,KAAK,CAAC,GAAG,IAAI,GAAG,CAAC,GAAGrD,OAAO,+DAA+D,CAAC;IACxP;IACA,IAAI,OAAOA,UAAU,YAAYA,UAAU,QAAQC,MAAMC,OAAO,CAACF,QAAQ;QACvE,MAAM,IAAInB,WAAW,kBAAkB,GAAG2B,WAAW,UAAU,EAAEa,KAAK,uEAAuE,CAAC;IAChJ;IACA,MAAMiC,QAAQtD;IACd,IAAI,SAASsD,OAAO;QAClB,MAAMzC,UAAUnB,OAAOC,IAAI,CAAC2D,OAAO1D,MAAM,CAAC,CAACC,IAAMA,MAAM;QACvD,IAAIgB,QAAQV,MAAM,GAAG,GAAG;YACtB,MAAM,IAAItB,WAAW,kBAAkB,GAAG2B,WAAW,UAAU,EAAEa,KAAK,oBAAoB,EAAER,QAAQC,IAAI,CAAC,MAAM,gCAAgC,CAAC;QAClJ;QACA,IAAI,OAAOwC,MAAMC,GAAG,KAAK,YAAYD,MAAMC,GAAG,CAACtC,IAAI,OAAO,IAAI;YAC5D,MAAM,IAAIpC,WAAW,kBAAkB,GAAG2B,WAAW,UAAU,EAAEa,KAAK,+BAA+B,CAAC;QACxG;QACA;IACF;IACA,IAAI,YAAYiC,OAAO;QACrB,MAAMzC,UAAUnB,OAAOC,IAAI,CAAC2D,OAAO1D,MAAM,CAAC,CAACC,IAAM,CAACN,kBAAkBO,GAAG,CAACD;QACxE,IAAIgB,QAAQV,MAAM,GAAG,GAAG;YACtB,MAAM,IAAItB,WAAW,kBAAkB,GAAG2B,WAAW,UAAU,EAAEa,KAAK,oBAAoB,EAAER,QAAQC,IAAI,CAAC,MAAM,iEAAiE,CAAC;QACnL;QACA,4EAA4E;QAC5E,IAAI,OAAOwC,MAAME,MAAM,KAAK,YAAYF,MAAME,MAAM,CAACvC,IAAI,OAAO,IAAI;YAClE,MAAM,IAAIpC,WAAW,kBAAkB,GAAG2B,WAAW,UAAU,EAAEa,KAAK,8BAA8B,CAAC;QACvG;QACA,IAAIiC,MAAMR,MAAM,KAAK9B,aAAc,CAAA,OAAOsC,MAAMR,MAAM,KAAK,YAAYQ,MAAMR,MAAM,CAAC3C,MAAM,KAAK,CAAA,GAAI;YACjG,MAAM,IAAItB,WAAW,kBAAkB,GAAG2B,WAAW,UAAU,EAAEa,KAAK,6BAA6B,CAAC;QACtG;QACA,IAAIiC,MAAM5C,OAAO,KAAKM,aAAa,CAACjB,sBAAsBuD,MAAM5C,OAAO,GAAG;YACxE,MAAM,IAAI7B,WAAW,kBAAkB,GAAG2B,WAAW,UAAU,EAAEa,KAAK,kDAAkD,CAAC;QAC3H;QACA,IAAIiC,MAAMP,OAAO,KAAK/B,aAAa,CAACjB,sBAAsBuD,MAAMP,OAAO,GAAG;YACxE,MAAM,IAAIlE,WAAW,kBAAkB,GAAG2B,WAAW,UAAU,EAAEa,KAAK,kDAAkD,CAAC;QAC3H;QACA,IAAIiC,MAAML,KAAK,KAAKjC,aAAa,OAAOsC,MAAML,KAAK,KAAK,UAAU;YAChE,MAAM,IAAIpE,WAAW,kBAAkB,GAAG2B,WAAW,UAAU,EAAEa,KAAK,qCAAqC,CAAC;QAC9G;QACA,IAAIiC,MAAMzD,CAAC,KAAKmB,aAAc,CAAA,OAAOsC,MAAMzD,CAAC,KAAK,YAAY,CAACsC,OAAOC,SAAS,CAACkB,MAAMzD,CAAC,KAAKyD,MAAMzD,CAAC,IAAI,CAAA,GAAI;YACxG,MAAM,IAAIhB,WAAW,kBAAkB,GAAG2B,WAAW,UAAU,EAAEa,KAAK,6BAA6B,CAAC;QACtG;QACA;IACF;IACA,MAAM,IAAIxC,WAAW,kBAAkB,GAAG2B,WAAW,UAAU,EAAEa,KAAK,uEAAuE,CAAC;AAChJ;AAEA,OAAO,SAASoC,eAAelD,MAAe,EAAEC,UAAkB;IAChE,IAAI,OAAOD,WAAW,YAAYA,WAAW,QAAQN,MAAMC,OAAO,CAACK,SAAS;QAC1E,MAAM,IAAI1B,WAAW,kBAAkB,GAAG2B,WAAW,8BAA8B,CAAC;IACtF;IACA,MAAMf,MAAMc;IAEZ,yFAAyF;IACzF,IAAId,IAAIiE,MAAM,KAAK1C,WAAW;QAC5B,MAAM,IAAInC,WAAW,kBAAkB,GAAG2B,WAAW,yHAAyH,CAAC;IACjL;IAEA,MAAMmD,UAAUlE,IAAIkE,OAAO;IAC3B,IAAI,CAACA,WAAW,OAAOA,YAAY,YAAY1D,MAAMC,OAAO,CAACyD,YAAYjE,OAAOC,IAAI,CAACgE,SAASxD,MAAM,KAAK,GAAG;QAC1G,MAAM,IAAItB,WAAW,kBAAkB,GAAG2B,WAAW,kGAAkG,CAAC;IAC1J;IACA,KAAK,MAAM,CAACa,MAAMrB,MAAM,IAAIN,OAAO8B,OAAO,CAACmC,SAAU;QACnDd,eAAexB,MAAMrB,OAAOQ;IAC9B;IACA,IAAImD,QAAQC,OAAO,KAAK5C,WAAW;QACjC,MAAM,IAAInC,WAAW,kBAAkB,GAAG2B,WAAW,yCAAyC,CAAC;IACjG;IAEA,IAAIf,IAAIoE,OAAO,KAAK7C,WAAWvB,IAAIoE,OAAO,GAAG,CAAC;IAC9C,IAAI,OAAOpE,IAAIoE,OAAO,KAAK,YAAYpE,IAAIoE,OAAO,KAAK,QAAQ5D,MAAMC,OAAO,CAACT,IAAIoE,OAAO,GAAG;QACzF,MAAM,IAAIhF,WAAW,kBAAkB,GAAG2B,WAAW,4DAA4D,CAAC;IACpH;IACA,MAAMqD,UAAUpE,IAAIoE,OAAO;IAC3B,KAAK,MAAM,CAACxC,MAAMrB,MAAM,IAAIN,OAAO8B,OAAO,CAACqC,SAAU;QACnDX,mBAAmB7B,MAAMrB,OAAOQ;IAClC;IAEA,IAAIf,IAAIqE,QAAQ,KAAK9C,WAAW;QAC9BM,sBAAsB7B,IAAIqE,QAAQ,EAAEtD;IACtC;IACA,IAAIf,IAAIiC,KAAK,KAAKV,WAAW;QAC3BS,mBAAmBhC,IAAIiC,KAAK,EAAElB;IAChC;IACA,IAAIf,IAAIsE,OAAO,KAAK/C,WAAW;QAC7BL,qBAAqBlB,IAAIsE,OAAO,EAAEvD;IACpC;IACA,IAAIf,IAAIuE,KAAK,KAAKhD,WAAW;QAC3BE,iBAAiBzB,IAAIuE,KAAK,EAAExD;IAC9B;IAEA,8FAA8F;IAC9F,IAAIf,IAAIiC,KAAK,KAAKV,WAAW;QAC3B,KAAK,MAAM,CAACK,MAAMrB,MAAM,IAAIN,OAAO8B,OAAO,CAACmC,SAAU;YACnD,MAAMX,UAAU,AAAChD,MAAkCgD,OAAO;YAC1D,IAAIA,WAAW,aAAaA,SAAS;gBACnC,MAAM,IAAInE,WAAW,kBAAkB,GAAG2B,WAAW,UAAU,EAAEa,KAAK,yGAAyG,EAAEA,KAAK,QAAQ,CAAC;YACjM;QACF;IACF;IAEA,OAAO5B;AACT"}

package/dist/esm/embed/cohere.js

@@ -25,8 +25,8 @@ export async function cohereProvider(model, url, keyEnv) {
         });
         if (!res.ok) throw new SenseError('EMBED_MODEL', `${base}/v2/embed -> HTTP ${res.status}`);
         const body = await res.json();
-        // Cohere returns one vector per text in order, with no index field to re-sort by; the
-        // count is what a caller mapping positionally still has to be able to trust.
+        // Cohere returns one vector per text in order, with no index field to re-sort by; a caller
+        // maps them positionally, so the count below must be trustworthy.
         const vectors = body.embeddings.float;
         if (vectors.length !== texts.length) throw new SenseError('EMBED_MODEL', `${base}/v2/embed returned ${vectors.length} embeddings for ${texts.length} inputs`);
         return vectors.map((v)=>Float32Array.from(v));

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

@@ -1 +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"}
+{"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 ${res.status}`);\n    const body = (await res.json()) as { embeddings: { float: number[][] } };\n    // Cohere returns one vector per text in order, with no index field to re-sort by; a caller\n    // maps them positionally, so the count below must be trustworthy.\n    const vectors = body.embeddings.float;\n    if (vectors.length !== texts.length) throw new SenseError('EMBED_MODEL', `${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,2FAA2F;QAC3F,kEAAkE;QAClE,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/distribution.js

@@ -1,4 +1,4 @@
-import { getMeta, setMeta } from '../store/meta.js';
+import { getMeta, setMeta } from '../store/shared.js';
 // Split out of langfit.ts so reading the persisted counts (status.ts's only need) never pulls in
 // franc-min; only checkLanguageFit's classify path needs the classifier itself.
 const META_KEY = 'embed_languages';

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

@@ -1 +1 @@
-{"version":3,"sources":["/Users/kevin/Dev/OpenSource/ai/sensemaking/src/embed/distribution.ts"],"sourcesContent":["import { getMeta, setMeta } from '../store/meta.ts';\nimport type { Store } from '../store/types.ts';\n\n// Split out of langfit.ts so reading the persisted counts (status.ts's only need) never pulls in\n// franc-min; only checkLanguageFit's classify path needs the classifier itself.\nconst META_KEY = 'embed_languages';\n\nexport type LangCounts = Record<string, number>;\n\nexport async function languageDistribution(store: Store): Promise<LangCounts | undefined> {\n  const raw = await getMeta(store, META_KEY);\n  return raw ? JSON.parse(raw) : undefined;\n}\n\nexport async function saveLanguageDistribution(store: Store, counts: LangCounts): Promise<void> {\n  await setMeta(store, META_KEY, JSON.stringify(counts));\n}\n"],"names":["getMeta","setMeta","META_KEY","languageDistribution","store","raw","JSON","parse","undefined","saveLanguageDistribution","counts","stringify"],"mappings":"AAAA,SAASA,OAAO,EAAEC,OAAO,QAAQ,mBAAmB;AAGpD,iGAAiG;AACjG,gFAAgF;AAChF,MAAMC,WAAW;AAIjB,OAAO,eAAeC,qBAAqBC,KAAY;IACrD,MAAMC,MAAM,MAAML,QAAQI,OAAOF;IACjC,OAAOG,MAAMC,KAAKC,KAAK,CAACF,OAAOG;AACjC;AAEA,OAAO,eAAeC,yBAAyBL,KAAY,EAAEM,MAAkB;IAC7E,MAAMT,QAAQG,OAAOF,UAAUI,KAAKK,SAAS,CAACD;AAChD"}
+{"version":3,"sources":["/Users/kevin/Dev/OpenSource/ai/sensemaking/src/embed/distribution.ts"],"sourcesContent":["import { getMeta, setMeta } from '../store/shared.ts';\nimport type { Store } from '../store/types.ts';\n\n// Split out of langfit.ts so reading the persisted counts (status.ts's only need) never pulls in\n// franc-min; only checkLanguageFit's classify path needs the classifier itself.\nconst META_KEY = 'embed_languages';\n\nexport type LangCounts = Record<string, number>;\n\nexport async function languageDistribution(store: Store): Promise<LangCounts | undefined> {\n  const raw = await getMeta(store, META_KEY);\n  return raw ? JSON.parse(raw) : undefined;\n}\n\nexport async function saveLanguageDistribution(store: Store, counts: LangCounts): Promise<void> {\n  await setMeta(store, META_KEY, JSON.stringify(counts));\n}\n"],"names":["getMeta","setMeta","META_KEY","languageDistribution","store","raw","JSON","parse","undefined","saveLanguageDistribution","counts","stringify"],"mappings":"AAAA,SAASA,OAAO,EAAEC,OAAO,QAAQ,qBAAqB;AAGtD,iGAAiG;AACjG,gFAAgF;AAChF,MAAMC,WAAW;AAIjB,OAAO,eAAeC,qBAAqBC,KAAY;IACrD,MAAMC,MAAM,MAAML,QAAQI,OAAOF;IACjC,OAAOG,MAAMC,KAAKC,KAAK,CAACF,OAAOG;AACjC;AAEA,OAAO,eAAeC,yBAAyBL,KAAY,EAAEM,MAAkB;IAC7E,MAAMT,QAAQG,OAAOF,UAAUI,KAAKK,SAAS,CAACD;AAChD"}

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

@@ -0,0 +1,3 @@
+export declare function stashChunkText(key: object, texts: Map<string, string[]>): void;
+export declare function rekeyChunkText(from: object, to: object): void;
+export declare function takeChunkText(key: object): Map<string, string[]> | undefined;

package/dist/esm/embed/handoff.js

@@ -0,0 +1,21 @@
+// Chunk text from reconcile to embed time when both run in one process, so the text is neither
+// re-derived nor persisted. Keyed by the store it was stashed against, so an entry dies with it.
+const HANDOFF = new WeakMap();
+export function stashChunkText(key, texts) {
+    if (texts.size > 0) HANDOFF.set(key, texts);
+}
+// Moves an entry from one key to another: reconcile runs before the store object exists, so it
+// stashes against the connection and open() re-keys onto the store it then builds.
+export function rekeyChunkText(from, to) {
+    const texts = HANDOFF.get(from);
+    if (!texts) return;
+    HANDOFF.delete(from);
+    HANDOFF.set(to, texts);
+}
+// Read once and drop: the text is consumed by the embed pass that follows reconcile, and holding
+// it past that would pin the tree's prose in memory for the life of the store.
+export function takeChunkText(key) {
+    const texts = HANDOFF.get(key);
+    HANDOFF.delete(key);
+    return texts;
+}

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

@@ -0,0 +1 @@
+{"version":3,"sources":["/Users/kevin/Dev/OpenSource/ai/sensemaking/src/embed/handoff.ts"],"sourcesContent":["// Chunk text from reconcile to embed time when both run in one process, so the text is neither\n// re-derived nor persisted. Keyed by the store it was stashed against, so an entry dies with it.\nconst HANDOFF = new WeakMap<object, Map<string, string[]>>();\n\nexport function stashChunkText(key: object, texts: Map<string, string[]>): void {\n  if (texts.size > 0) HANDOFF.set(key, texts);\n}\n\n// Moves an entry from one key to another: reconcile runs before the store object exists, so it\n// stashes against the connection and open() re-keys onto the store it then builds.\nexport function rekeyChunkText(from: object, to: object): void {\n  const texts = HANDOFF.get(from);\n  if (!texts) return;\n  HANDOFF.delete(from);\n  HANDOFF.set(to, texts);\n}\n\n// Read once and drop: the text is consumed by the embed pass that follows reconcile, and holding\n// it past that would pin the tree's prose in memory for the life of the store.\nexport function takeChunkText(key: object): Map<string, string[]> | undefined {\n  const texts = HANDOFF.get(key);\n  HANDOFF.delete(key);\n  return texts;\n}\n"],"names":["HANDOFF","WeakMap","stashChunkText","key","texts","size","set","rekeyChunkText","from","to","get","delete","takeChunkText"],"mappings":"AAAA,+FAA+F;AAC/F,iGAAiG;AACjG,MAAMA,UAAU,IAAIC;AAEpB,OAAO,SAASC,eAAeC,GAAW,EAAEC,KAA4B;IACtE,IAAIA,MAAMC,IAAI,GAAG,GAAGL,QAAQM,GAAG,CAACH,KAAKC;AACvC;AAEA,+FAA+F;AAC/F,mFAAmF;AACnF,OAAO,SAASG,eAAeC,IAAY,EAAEC,EAAU;IACrD,MAAML,QAAQJ,QAAQU,GAAG,CAACF;IAC1B,IAAI,CAACJ,OAAO;IACZJ,QAAQW,MAAM,CAACH;IACfR,QAAQM,GAAG,CAACG,IAAIL;AAClB;AAEA,iGAAiG;AACjG,+EAA+E;AAC/E,OAAO,SAASQ,cAAcT,GAAW;IACvC,MAAMC,QAAQJ,QAAQU,GAAG,CAACP;IAC1BH,QAAQW,MAAM,CAACR;IACf,OAAOC;AACT"}

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

@@ -1,13 +1,14 @@
 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;
+export declare const DEFAULT_MODELS_ROOT: string;
+export declare function refsMainPath(model: string, root?: string): string;
+export declare function readRef(model: string, root?: string): string | undefined;
+export declare function writeRef(model: string, sha: string, root?: string): void;
+export declare function snapshotDir(model: string, sha: string, root?: string): string;
+export declare function languagesPath(model: string, root?: string): string;
+export declare function readLanguages(model: string, root?: string): string[] | undefined;
+export declare function writeLanguages(model: string, languages: string[], root?: string): void;
+export declare function modelDir(model: string, root?: string): string;
+export declare function hasModelFiles(model: string, root?: string): boolean;
+export declare function modelIdentity(model: string, root?: string): string | undefined;

package/dist/esm/embed/identity.js

@@ -16,40 +16,43 @@ const HF_MODEL_ID = /^[A-Za-z0-9][\w.-]*\/[A-Za-z0-9][\w.-]*$/;
 export function isDownloadable(model) {
     return HF_MODEL_ID.test(model);
 }
-function repoDir(model) {
+// The one production default; tests pass their own scratch root through the same parameter
+// rather than seeding fixture ids into this machine cache.
+export const DEFAULT_MODELS_ROOT = join(homedir(), '.sense', 'models');
+function repoDir(model, root = DEFAULT_MODELS_ROOT) {
     const [org, name] = model.split('/');
-    return join(homedir(), '.sense', 'models', `models--${org}--${name}`);
+    return join(root, `models--${org}--${name}`);
 }
-export function refsMainPath(model) {
-    return join(repoDir(model), 'refs', 'main');
+export function refsMainPath(model, root = DEFAULT_MODELS_ROOT) {
+    return join(repoDir(model, root), '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);
+export function readRef(model, root = DEFAULT_MODELS_ROOT) {
+    const p = refsMainPath(model, root);
     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'), {
+export function writeRef(model, sha, root = DEFAULT_MODELS_ROOT) {
+    mkdirSync(join(repoDir(model, root), 'refs'), {
         recursive: true
     });
-    writeFileAtomic(refsMainPath(model), sha);
+    writeFileAtomic(refsMainPath(model, root), sha);
 }
-export function snapshotDir(model, sha) {
-    return join(repoDir(model), 'snapshots', sha);
+export function snapshotDir(model, sha, root = DEFAULT_MODELS_ROOT) {
+    return join(repoDir(model, root), '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');
+export function languagesPath(model, root = DEFAULT_MODELS_ROOT) {
+    return join(repoDir(model, root), '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);
+export function readLanguages(model, root = DEFAULT_MODELS_ROOT) {
+    const p = languagesPath(model, root);
     if (!existsSync(p)) return undefined;
     try {
         return JSON.parse(readFileSync(p, 'utf8'));
@@ -57,29 +60,29 @@ export function readLanguages(model) {
         return undefined;
     }
 }
-export function writeLanguages(model, languages) {
-    mkdirSync(repoDir(model), {
+export function writeLanguages(model, languages, root = DEFAULT_MODELS_ROOT) {
+    mkdirSync(repoDir(model, root), {
         recursive: true
     });
-    writeFileAtomic(languagesPath(model), JSON.stringify(languages));
+    writeFileAtomic(languagesPath(model, root), 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) {
+export function modelDir(model, root = DEFAULT_MODELS_ROOT) {
     if (!isDownloadable(model)) return model;
-    const sha = readRef(model);
-    return sha ? snapshotDir(model, sha) : repoDir(model);
+    const sha = readRef(model, root);
+    return sha ? snapshotDir(model, sha, root) : repoDir(model, root);
 }
 // Both files, so an interrupted download reads as absent and the next one resumes it.
-export function hasModelFiles(model) {
-    const dir = modelDir(model);
+export function hasModelFiles(model, root = DEFAULT_MODELS_ROOT) {
+    const dir = modelDir(model, root);
     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);
+export function modelIdentity(model, root = DEFAULT_MODELS_ROOT) {
+    if (isDownloadable(model)) return readRef(model, root);
+    const dir = modelDir(model, root);
     const parts = [];
     for (const file of MODEL_FILES){
         try {

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

@@ -1 +1 @@
-{"version":3,"sources":["/Users/kevin/Dev/OpenSource/ai/sensemaking/src/embed/identity.ts"],"sourcesContent":["import { existsSync, mkdirSync, readFileSync, statSync } from 'node:fs';\nimport { homedir } from 'node:os';\nimport { join } from 'node:path';\nimport { writeFileAtomic } from '../lib/atomic-write.ts';\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  writeFileAtomic(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  writeFileAtomic(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","homedir","join","writeFileAtomic","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,QAAQ,UAAU;AACxE,SAASC,OAAO,QAAQ,UAAU;AAClC,SAASC,IAAI,QAAQ,YAAY;AACjC,SAASC,eAAe,QAAQ,yBAAyB;AAEzD,iFAAiF;AACjF,2FAA2F;AAE3F,OAAO,MAAMC,cAAc;IAAC;IAAqB;CAAiB,CAAC;AACnE,8EAA8E;AAC9E,OAAO,MAAMC,kBAAkBD,YAAYF,IAAI,CAAC,SAAS;AAEzD,+FAA+F;AAC/F,sCAAsC;AACtC,MAAMI,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,OAAOX,KAAKD,WAAW,UAAU,UAAU,CAAC,QAAQ,EAAEU,IAAI,EAAE,EAAEC,MAAM;AACtE;AAEA,OAAO,SAASE,aAAaN,KAAa;IACxC,OAAON,KAAKQ,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,UAAUI,KAAKQ,QAAQF,QAAQ,SAAS;QAAEc,WAAW;IAAK;IAC1DnB,gBAAgBW,aAAaN,QAAQU;AACvC;AAEA,OAAO,SAASK,YAAYf,KAAa,EAAEU,GAAW;IACpD,OAAOhB,KAAKQ,QAAQF,QAAQ,aAAaU;AAC3C;AAEA,wFAAwF;AACxF,yBAAyB;AACzB,OAAO,SAASM,cAAchB,KAAa;IACzC,OAAON,KAAKQ,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;IAC5CnB,gBAAgBqB,cAAchB,QAAQkB,KAAKI,SAAS,CAACD;AACvD;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,WAAWK,KAAK+B,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,SAASE,KAAK+B,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,MAAMnC,IAAI,CAAC;AACpB"}
+{"version":3,"sources":["/Users/kevin/Dev/OpenSource/ai/sensemaking/src/embed/identity.ts"],"sourcesContent":["import { existsSync, mkdirSync, readFileSync, statSync } from 'node:fs';\nimport { homedir } from 'node:os';\nimport { join } from 'node:path';\nimport { writeFileAtomic } from '../lib/atomic-write.ts';\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\n// The one production default; tests pass their own scratch root through the same parameter\n// rather than seeding fixture ids into this machine cache.\nexport const DEFAULT_MODELS_ROOT = join(homedir(), '.sense', 'models');\n\nfunction repoDir(model: string, root: string = DEFAULT_MODELS_ROOT): string {\n  const [org, name] = model.split('/');\n  return join(root, `models--${org}--${name}`);\n}\n\nexport function refsMainPath(model: string, root: string = DEFAULT_MODELS_ROOT): string {\n  return join(repoDir(model, root), '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, root: string = DEFAULT_MODELS_ROOT): string | undefined {\n  const p = refsMainPath(model, root);\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, root: string = DEFAULT_MODELS_ROOT): void {\n  mkdirSync(join(repoDir(model, root), 'refs'), { recursive: true });\n  writeFileAtomic(refsMainPath(model, root), sha);\n}\n\nexport function snapshotDir(model: string, sha: string, root: string = DEFAULT_MODELS_ROOT): string {\n  return join(repoDir(model, root), '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, root: string = DEFAULT_MODELS_ROOT): string {\n  return join(repoDir(model, root), '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, root: string = DEFAULT_MODELS_ROOT): string[] | undefined {\n  const p = languagesPath(model, root);\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[], root: string = DEFAULT_MODELS_ROOT): void {\n  mkdirSync(repoDir(model, root), { recursive: true });\n  writeFileAtomic(languagesPath(model, root), 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, root: string = DEFAULT_MODELS_ROOT): string {\n  if (!isDownloadable(model)) return model;\n  const sha = readRef(model, root);\n  return sha ? snapshotDir(model, sha, root) : repoDir(model, root);\n}\n\n// Both files, so an interrupted download reads as absent and the next one resumes it.\nexport function hasModelFiles(model: string, root: string = DEFAULT_MODELS_ROOT): boolean {\n  const dir = modelDir(model, root);\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, root: string = DEFAULT_MODELS_ROOT): string | undefined {\n  if (isDownloadable(model)) return readRef(model, root);\n  const dir = modelDir(model, root);\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","homedir","join","writeFileAtomic","MODEL_FILES","MODEL_FILENAMES","HF_MODEL_ID","isDownloadable","model","test","DEFAULT_MODELS_ROOT","repoDir","root","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,QAAQ,UAAU;AACxE,SAASC,OAAO,QAAQ,UAAU;AAClC,SAASC,IAAI,QAAQ,YAAY;AACjC,SAASC,eAAe,QAAQ,yBAAyB;AAEzD,iFAAiF;AACjF,2FAA2F;AAE3F,OAAO,MAAMC,cAAc;IAAC;IAAqB;CAAiB,CAAC;AACnE,8EAA8E;AAC9E,OAAO,MAAMC,kBAAkBD,YAAYF,IAAI,CAAC,SAAS;AAEzD,+FAA+F;AAC/F,sCAAsC;AACtC,MAAMI,cAAc;AAEpB,OAAO,SAASC,eAAeC,KAAa;IAC1C,OAAOF,YAAYG,IAAI,CAACD;AAC1B;AAEA,2FAA2F;AAC3F,2DAA2D;AAC3D,OAAO,MAAME,sBAAsBR,KAAKD,WAAW,UAAU,UAAU;AAEvE,SAASU,QAAQH,KAAa,EAAEI,OAAeF,mBAAmB;IAChE,MAAM,CAACG,KAAKC,KAAK,GAAGN,MAAMO,KAAK,CAAC;IAChC,OAAOb,KAAKU,MAAM,CAAC,QAAQ,EAAEC,IAAI,EAAE,EAAEC,MAAM;AAC7C;AAEA,OAAO,SAASE,aAAaR,KAAa,EAAEI,OAAeF,mBAAmB;IAC5E,OAAOR,KAAKS,QAAQH,OAAOI,OAAO,QAAQ;AAC5C;AAEA,8FAA8F;AAC9F,4DAA4D;AAC5D,OAAO,SAASK,QAAQT,KAAa,EAAEI,OAAeF,mBAAmB;IACvE,MAAMQ,IAAIF,aAAaR,OAAOI;IAC9B,IAAI,CAACf,WAAWqB,IAAI,OAAOC;IAC3B,MAAMC,MAAMrB,aAAamB,GAAG,QAAQG,IAAI;IACxC,OAAOD,IAAIE,MAAM,GAAG,IAAIF,MAAMD;AAChC;AAEA,4FAA4F;AAC5F,OAAO,SAASI,SAASf,KAAa,EAAEY,GAAW,EAAER,OAAeF,mBAAmB;IACrFZ,UAAUI,KAAKS,QAAQH,OAAOI,OAAO,SAAS;QAAEY,WAAW;IAAK;IAChErB,gBAAgBa,aAAaR,OAAOI,OAAOQ;AAC7C;AAEA,OAAO,SAASK,YAAYjB,KAAa,EAAEY,GAAW,EAAER,OAAeF,mBAAmB;IACxF,OAAOR,KAAKS,QAAQH,OAAOI,OAAO,aAAaQ;AACjD;AAEA,wFAAwF;AACxF,yBAAyB;AACzB,OAAO,SAASM,cAAclB,KAAa,EAAEI,OAAeF,mBAAmB;IAC7E,OAAOR,KAAKS,QAAQH,OAAOI,OAAO;AACpC;AAEA,yFAAyF;AACzF,6FAA6F;AAC7F,OAAO,SAASe,cAAcnB,KAAa,EAAEI,OAAeF,mBAAmB;IAC7E,MAAMQ,IAAIQ,cAAclB,OAAOI;IAC/B,IAAI,CAACf,WAAWqB,IAAI,OAAOC;IAC3B,IAAI;QACF,OAAOS,KAAKC,KAAK,CAAC9B,aAAamB,GAAG;IACpC,EAAE,OAAM;QACN,OAAOC;IACT;AACF;AAEA,OAAO,SAASW,eAAetB,KAAa,EAAEuB,SAAmB,EAAEnB,OAAeF,mBAAmB;IACnGZ,UAAUa,QAAQH,OAAOI,OAAO;QAAEY,WAAW;IAAK;IAClDrB,gBAAgBuB,cAAclB,OAAOI,OAAOgB,KAAKI,SAAS,CAACD;AAC7D;AAEA,8FAA8F;AAC9F,4FAA4F;AAC5F,OAAO,SAASE,SAASzB,KAAa,EAAEI,OAAeF,mBAAmB;IACxE,IAAI,CAACH,eAAeC,QAAQ,OAAOA;IACnC,MAAMY,MAAMH,QAAQT,OAAOI;IAC3B,OAAOQ,MAAMK,YAAYjB,OAAOY,KAAKR,QAAQD,QAAQH,OAAOI;AAC9D;AAEA,sFAAsF;AACtF,OAAO,SAASsB,cAAc1B,KAAa,EAAEI,OAAeF,mBAAmB;IAC7E,MAAMyB,MAAMF,SAASzB,OAAOI;IAC5B,OAAOR,YAAYgC,KAAK,CAAC,CAACC,OAASxC,WAAWK,KAAKiC,KAAKE;AAC1D;AAEA,oFAAoF;AACpF,2EAA2E;AAC3E,OAAO,SAASC,cAAc9B,KAAa,EAAEI,OAAeF,mBAAmB;IAC7E,IAAIH,eAAeC,QAAQ,OAAOS,QAAQT,OAAOI;IACjD,MAAMuB,MAAMF,SAASzB,OAAOI;IAC5B,MAAM2B,QAAkB,EAAE;IAC1B,KAAK,MAAMF,QAAQjC,YAAa;QAC9B,IAAI;YACF,MAAMoC,KAAKxC,SAASE,KAAKiC,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,MAAMrC,IAAI,CAAC;AACpB"}

package/dist/esm/embed/langfit.js

@@ -8,9 +8,8 @@ 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;
-// franc-min is a 148K static import; a top-level import put it on every command's module graph
-// regardless of whether the semantic path ran, shipped as the 0.17.0 regression. Deferred to a
-// synchronous require inside checkLanguageFit so only the embed path pays for it.
+// franc-min is a 148K static import; a top-level import would put it on every command's module
+// graph regardless of whether the semantic path ran, so it's deferred to a synchronous require.
 const _require = typeof require === 'undefined' ? Module.createRequire(import.meta.url) : require;
 function classify(franc, text) {
     const code = franc(text.slice(0, SAMPLE_CHARS));

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

@@ -1 +1 @@
-{"version":3,"sources":["/Users/kevin/Dev/OpenSource/ai/sensemaking/src/embed/langfit.ts"],"sourcesContent":["import Module from 'node:module';\nimport { SenseError } from '../errors.ts';\nimport type { Store } from '../store/types.ts';\nimport { languageDistribution, saveLanguageDistribution } from './distribution.ts';\nimport { toIso3 } from './languages.ts';\nimport type { EmbedProvider } from './types.ts';\n\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\n// franc-min is a 148K static import; a top-level import put it on every command's module graph\n// regardless of whether the semantic path ran, shipped as the 0.17.0 regression. Deferred to a\n// synchronous require inside checkLanguageFit so only the embed path pays for it.\nconst _require = typeof require === 'undefined' ? Module.createRequire(import.meta.url) : require;\n\nfunction classify(franc: typeof import('franc-min')['franc'], 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(store: Store, provider: EmbedProvider, texts: string[]): Promise<void> {\n  if (texts.length === 0) return;\n  const { franc } = _require('franc-min') as typeof import('franc-min');\n  const persisted = (await languageDistribution(store)) ?? {};\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  await saveLanguageDistribution(store, merged);\n}\n"],"names":["Module","SenseError","languageDistribution","saveLanguageDistribution","toIso3","SAMPLE_CHARS","MIN_CLASSIFIED","_require","require","createRequire","url","classify","franc","text","code","slice","undefined","checkLanguageFit","store","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"],"mappings":"AAAA,OAAOA,YAAY,cAAc;AACjC,SAASC,UAAU,QAAQ,eAAe;AAE1C,SAASC,oBAAoB,EAAEC,wBAAwB,QAAQ,oBAAoB;AACnF,SAASC,MAAM,QAAQ,iBAAiB;AAGxC,+FAA+F;AAC/F,8FAA8F;AAC9F,MAAMC,eAAe;AACrB,gGAAgG;AAChG,4EAA4E;AAC5E,MAAMC,iBAAiB;AAEvB,+FAA+F;AAC/F,+FAA+F;AAC/F,kFAAkF;AAClF,MAAMC,WAAW,OAAOC,YAAY,cAAcR,OAAOS,aAAa,CAAC,YAAYC,GAAG,IAAIF;AAE1F,SAASG,SAASC,KAA0C,EAAEC,IAAY;IACxE,MAAMC,OAAOF,MAAMC,KAAKE,KAAK,CAAC,GAAGV;IACjC,OAAOS,SAAS,QAAQE,YAAYF;AACtC;AAEA,+FAA+F;AAC/F,+EAA+E;AAC/E,OAAO,eAAeG,iBAAiBC,KAAY,EAAEC,QAAuB,EAAEC,KAAe;QAGxE;IAFnB,IAAIA,MAAMC,MAAM,KAAK,GAAG;IACxB,MAAM,EAAET,KAAK,EAAE,GAAGL,SAAS;IAC3B,MAAMe,aAAa,OAAA,MAAMpB,qBAAqBgB,oBAA3B,kBAAA,OAAsC,CAAC;IAC1D,MAAMK,SAAS;QAAE,GAAGD,SAAS;IAAC;IAC9B,KAAK,MAAMT,QAAQO,MAAO;YAEEG;QAD1B,MAAMT,OAAOH,SAASC,OAAOC;QAC7B,IAAIC,MAAMS,MAAM,CAACT,KAAK,GAAG,EAACS,eAAAA,MAAM,CAACT,KAAK,cAAZS,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,CAAC7B;QAChD,MAAM,CAAC8B,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,IAAIvB,WACR,wBACA,CAAC,YAAY,EAAEkB,SAASuB,EAAE,CAAC,qBAAqB,EAAEvB,SAASW,SAAS,CAACa,IAAI,CAAC,MAAM,OAAO,EAAEJ,IAAI,qCAAqC,EAAEL,aAAa,0JAA0J,CAAC;QAEhT;IACF;IACA,MAAM/B,yBAAyBe,OAAOK;AACxC"}
+{"version":3,"sources":["/Users/kevin/Dev/OpenSource/ai/sensemaking/src/embed/langfit.ts"],"sourcesContent":["import Module from 'node:module';\nimport { SenseError } from '../errors.ts';\nimport type { Store } from '../store/types.ts';\nimport { languageDistribution, saveLanguageDistribution } from './distribution.ts';\nimport { toIso3 } from './languages.ts';\nimport type { EmbedProvider } from './types.ts';\n\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\n// franc-min is a 148K static import; a top-level import would put it on every command's module\n// graph regardless of whether the semantic path ran, so it's deferred to a synchronous require.\nconst _require = typeof require === 'undefined' ? Module.createRequire(import.meta.url) : require;\n\nfunction classify(franc: typeof import('franc-min')['franc'], 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(store: Store, provider: EmbedProvider, texts: string[]): Promise<void> {\n  if (texts.length === 0) return;\n  const { franc } = _require('franc-min') as typeof import('franc-min');\n  const persisted = (await languageDistribution(store)) ?? {};\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  await saveLanguageDistribution(store, merged);\n}\n"],"names":["Module","SenseError","languageDistribution","saveLanguageDistribution","toIso3","SAMPLE_CHARS","MIN_CLASSIFIED","_require","require","createRequire","url","classify","franc","text","code","slice","undefined","checkLanguageFit","store","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"],"mappings":"AAAA,OAAOA,YAAY,cAAc;AACjC,SAASC,UAAU,QAAQ,eAAe;AAE1C,SAASC,oBAAoB,EAAEC,wBAAwB,QAAQ,oBAAoB;AACnF,SAASC,MAAM,QAAQ,iBAAiB;AAGxC,+FAA+F;AAC/F,8FAA8F;AAC9F,MAAMC,eAAe;AACrB,gGAAgG;AAChG,4EAA4E;AAC5E,MAAMC,iBAAiB;AAEvB,+FAA+F;AAC/F,gGAAgG;AAChG,MAAMC,WAAW,OAAOC,YAAY,cAAcR,OAAOS,aAAa,CAAC,YAAYC,GAAG,IAAIF;AAE1F,SAASG,SAASC,KAA0C,EAAEC,IAAY;IACxE,MAAMC,OAAOF,MAAMC,KAAKE,KAAK,CAAC,GAAGV;IACjC,OAAOS,SAAS,QAAQE,YAAYF;AACtC;AAEA,+FAA+F;AAC/F,+EAA+E;AAC/E,OAAO,eAAeG,iBAAiBC,KAAY,EAAEC,QAAuB,EAAEC,KAAe;QAGxE;IAFnB,IAAIA,MAAMC,MAAM,KAAK,GAAG;IACxB,MAAM,EAAET,KAAK,EAAE,GAAGL,SAAS;IAC3B,MAAMe,aAAa,OAAA,MAAMpB,qBAAqBgB,oBAA3B,kBAAA,OAAsC,CAAC;IAC1D,MAAMK,SAAS;QAAE,GAAGD,SAAS;IAAC;IAC9B,KAAK,MAAMT,QAAQO,MAAO;YAEEG;QAD1B,MAAMT,OAAOH,SAASC,OAAOC;QAC7B,IAAIC,MAAMS,MAAM,CAACT,KAAK,GAAG,EAACS,eAAAA,MAAM,CAACT,KAAK,cAAZS,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,CAAC7B;QAChD,MAAM,CAAC8B,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,IAAIvB,WACR,wBACA,CAAC,YAAY,EAAEkB,SAASuB,EAAE,CAAC,qBAAqB,EAAEvB,SAASW,SAAS,CAACa,IAAI,CAAC,MAAM,OAAO,EAAEJ,IAAI,qCAAqC,EAAEL,aAAa,0JAA0J,CAAC;QAEhT;IACF;IACA,MAAM/B,yBAAyBe,OAAOK;AACxC"}

package/dist/esm/embed/languages.js

@@ -1,7 +1,5 @@
-// 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.
+// HF model cards declare languages as ISO 639-1 (e.g. "zh"); franc-min reports ISO 639-3 (e.g.
+// "cmn"). Limited to franc-min's own 82-language table (https://github.com/wooorm/franc).
 const ISO_639_1_TO_3 = {
     ar: 'arb',
     am: 'amh',

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

@@ -1 +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'], // benchmark/reports/2026-08-27-embedding-model-selection.md\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"}
+{"version":3,"sources":["/Users/kevin/Dev/OpenSource/ai/sensemaking/src/embed/languages.ts"],"sourcesContent":["// HF model cards declare languages as ISO 639-1 (e.g. \"zh\"); franc-min reports ISO 639-3 (e.g.\n// \"cmn\"). Limited to franc-min's own 82-language table (https://github.com/wooorm/franc).\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'], // benchmark/reports/2026-08-27-embedding-model-selection.md\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,+FAA+F;AAC/F,0FAA0F;AAC1F,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"}

package/dist/esm/embed/openai.js

@@ -35,8 +35,8 @@ export async function openaiProvider(model, url, keyEnv) {
     const dims = (await post([
         'dimension probe'
     ]))[0].length;
-    // Symmetric for now: per-model prefixes (nomic's search_document/search_query) are a
-    // later workstream.
+    // embedQuery and embedDocuments are symmetric here; per-model prefixes like nomic's
+    // search_document/search_query are not applied.
     return {
         id: `openai:${base}:${model}`,
         dims,