sensemaking 0.23.0 → 0.24.0

package/dist/cjs/scan/frontmatter.js.map

@@ -1 +1 @@
-{"version":3,"sources":["/Users/kevin/Dev/OpenSource/ai/sensemaking/src/scan/frontmatter.ts"],"sourcesContent":["import { isCollection, parseDocument, visit } from 'yaml';\n\n// Frontmatter keys that would collide with table columns. Exported so reconcile.ts's upsert can tell\n// a feature-owned column (`_rank`) from a parsed one and leave it alone on reparse.\nexport const RESERVED_COLUMNS = new Set(['path', '_mtime', '_ctime', '_size', '_rank', '_parse_error', 'content', 'links', 'sections']);\n\n// Only BAD_SCALAR_START qualifies: YAML 1.2 reserves `@` and `` ` `` at the start of a plain scalar, so `aliases: [@handle]` can only mean [\"@handle\"].\n// Every other code has a second reading (a swallowed key, a dropped URL, a silently overwritten duplicate) that would write values nobody wrote.\nconst ACCEPTED_YAML_CODES = new Set(['BAD_SCALAR_START']);\n\nexport function normalizeText(value: unknown): string {\n  if (value === null || value === undefined) return '';\n  return String(value).replace(/\\s+/g, ' ').trim();\n}\n\n// SQLite's datetime() rejects a colonless offset (`-0800`) and a space separator, which ISO 8601\n// allows and producers emit. A rejected date is invisible, not excluded: every comparison is NULL.\nconst ISO_DATETIME = /^(\\d{4}-\\d{2}-\\d{2})[T ](\\d{2}:\\d{2}(?::\\d{2})?(?:\\.\\d+)?)(Z|[+-]\\d{2}(?::?\\d{2})?)?$/;\n\n// A value opening `YYYY-MM-DDT` was meant to be a datetime; prose never is. Reported when it\n// cannot be normalized, so a typo surfaces on the next crawl instead of at some later audit.\nconst MEANT_AS_DATETIME = /^\\d{4}-\\d{2}-\\d{2}[T ]\\d/;\n\nexport function looksLikeDatetime(value: string): boolean {\n  return MEANT_AS_DATETIME.test(value);\n}\n\n// Punctuation only, never a timezone conversion: the offset survives, so substr(d,1,10) is still\n// the local date. A shape that is not a real instant is left as written, and stays auditable.\nexport function normalizeDate(value: string): string {\n  const m = ISO_DATETIME.exec(value);\n  if (m === null) return value;\n  const [, date, time, zone] = m;\n  const digits = zone === undefined || zone === 'Z' ? '' : zone.replace(':', '');\n  const offset = digits === '' ? (zone ?? '') : digits.length === 3 ? `${digits}:00` : `${digits.slice(0, 3)}:${digits.slice(3)}`;\n  const normalized = `${date}T${time}${offset}`;\n  return Number.isNaN(Date.parse(normalized)) ? value : normalized;\n}\n\n// Storage class follows the YAML scalar. Booleans store as 1/0, so `WHERE flag = 1` matches\n// and `WHERE flag = 'true'` cannot; `map` prints observed types so the mismatch is visible.\nexport function mapValue(value: unknown): string | number | bigint | null {\n  if (value === null || value === undefined) return null;\n  if (typeof value === 'boolean') return BigInt(value ? 1 : 0);\n  if (typeof value === 'number') return Number.isSafeInteger(value) ? BigInt(value) : value;\n  if (typeof value === 'string') return normalizeDate(value);\n  return JSON.stringify(value);\n}\n\n// Exactly `s.split('\\n').length` without allocating the array: callers that need only the count\n// run it per file per feature on the parse hot path.\nexport function countLines(s: string): number {\n  let n = 1;\n  for (let i = 0; i < s.length; i++) if (s.charCodeAt(i) === 10) n++;\n  return n;\n}\n\n// The delimiter split is all this package used gray-matter for.\nexport function splitFrontmatter(raw: string): { fm: string | null; body: string } {\n  const open = raw.match(/^---\\r?\\n/);\n  if (!open) return { fm: null, body: raw };\n  const rest = raw.slice(open[0].length);\n  const close = rest.match(/^---\\r?(\\n|$)/m);\n  if (!close || close.index === undefined) return { fm: null, body: raw };\n  return { fm: rest.slice(0, close.index), body: rest.slice(close.index + close[0].length) };\n}\n\n// A well-formed document can hold a value nobody meant: `created: {{date}}` is valid YAML for a flow map used as a mapping key, so it raises no error and stores {\"{ date }\": null}.\n// No error code catches that; yaml's own warning lacks a path and fires once per document, so this reports it here instead, with the path.\nfunction warnStringifiedKeys(relPath: string, doc: ReturnType<typeof parseDocument>, warnings: string[]): void {\n  let found = false;\n  // Nested, not top level: `created: {{date}}` puts the collection key one level down, inside\n  // the flow map that `{{...}}` parses as.\n  visit(doc, {\n    Pair(_key, pair) {\n      if (!isCollection(pair.key)) return undefined;\n      found = true;\n      return visit.BREAK;\n    },\n  });\n  // One per file: a template repeats the same mistake on every field it stamps.\n  if (found) warnings.push(`warning: ${relPath} frontmatter has a key that is itself a list or mapping, stored as text; this is usually an unrendered template placeholder like {{date}}`);\n}\n\n// Anything past ACCEPTED_YAML_CODES is quarantined: no frontmatter columns, `_parse_error` carries the reason. Recovering it would write values nobody wrote, worse than absence since no query can see it; content, links and sections never touch frontmatter, so a broken note stays searchable.\n// yaml's message continues onto a source excerpt, so the first line is the sentence, minus the colon that introduced the part being dropped.\nfunction firstLine(message: string): string {\n  return message.split('\\n')[0].replace(/:\\s*$/, '');\n}\n\nexport function parseFrontmatter(relPath: string, fm: string, warnings: string[]): { data: Record<string, unknown>; parseError: string | null } {\n  // logLevel silences yaml's own pathless warnings; warnStringifiedKeys re-reports the one\n  // that carries information, with the file it came from.\n  const doc = parseDocument(fm, { logLevel: 'silent' });\n  const refused = doc.errors.filter((err) => !ACCEPTED_YAML_CODES.has(err.code));\n  if (refused.length > 0) {\n    const detail = refused.length > 1 ? ` (and ${refused.length - 1} more)` : '';\n    const parseError = `${firstLine(refused[0].message)}${detail}`;\n    warnings.push(`warning: ${relPath} frontmatter did not parse, so none of it is indexed: ${parseError}`);\n    return { data: {}, parseError };\n  }\n\n  let data: unknown;\n  try {\n    data = doc.toJS();\n  } catch (err) {\n    // Reaches here with doc.errors empty: `title: **Bold**` parses, then opens an alias on\n    // materialisation. An empty error list is not a successful parse.\n    const parseError = firstLine((err as Error).message);\n    warnings.push(`warning: ${relPath} frontmatter did not parse, so none of it is indexed: ${parseError}`);\n    return { data: {}, parseError };\n  }\n\n  if (data === null || data === undefined) return { data: {}, parseError: null };\n  if (typeof data !== 'object' || Array.isArray(data)) {\n    const parseError = 'frontmatter is not a key-value mapping';\n    warnings.push(`warning: ${relPath} ${parseError}; none of it is indexed`);\n    return { data: {}, parseError };\n  }\n  warnStringifiedKeys(relPath, doc, warnings);\n  return { data: data as Record<string, unknown>, parseError: null };\n}\n"],"names":["RESERVED_COLUMNS","countLines","looksLikeDatetime","mapValue","normalizeDate","normalizeText","parseFrontmatter","splitFrontmatter","Set","ACCEPTED_YAML_CODES","value","undefined","String","replace","trim","ISO_DATETIME","MEANT_AS_DATETIME","test","m","exec","date","time","zone","digits","offset","length","slice","normalized","Number","isNaN","Date","parse","BigInt","isSafeInteger","JSON","stringify","s","n","i","charCodeAt","raw","open","match","fm","body","rest","close","index","warnStringifiedKeys","relPath","doc","warnings","found","visit","Pair","_key","pair","isCollection","key","BREAK","push","firstLine","message","split","parseDocument","logLevel","refused","errors","filter","err","has","code","detail","parseError","data","toJS","Array","isArray"],"mappings":";;;;;;;;;;;QAIaA;eAAAA;;QA+CGC;eAAAA;;QA5BAC;eAAAA;;QAkBAC;eAAAA;;QAZAC;eAAAA;;QAnBAC;eAAAA;;QAgFAC;eAAAA;;QAhCAC;eAAAA;;;oBA1DmC;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;AAI5C,IAAMP,mBAAmB,IAAIQ,IAAI;IAAC;IAAQ;IAAU;IAAU;IAAS;IAAS;IAAgB;IAAW;IAAS;CAAW;AAEtI,wJAAwJ;AACxJ,iJAAiJ;AACjJ,IAAMC,sBAAsB,IAAID,IAAI;IAAC;CAAmB;AAEjD,SAASH,cAAcK,KAAc;IAC1C,IAAIA,UAAU,QAAQA,UAAUC,WAAW,OAAO;IAClD,OAAOC,OAAOF,OAAOG,OAAO,CAAC,QAAQ,KAAKC,IAAI;AAChD;AAEA,iGAAiG;AACjG,mGAAmG;AACnG,IAAMC,eAAe;AAErB,6FAA6F;AAC7F,6FAA6F;AAC7F,IAAMC,oBAAoB;AAEnB,SAASd,kBAAkBQ,KAAa;IAC7C,OAAOM,kBAAkBC,IAAI,CAACP;AAChC;AAIO,SAASN,cAAcM,KAAa;IACzC,IAAMQ,IAAIH,aAAaI,IAAI,CAACT;IAC5B,IAAIQ,MAAM,MAAM,OAAOR;IACvB,IAA6BQ,sBAAAA,OAApBE,OAAoBF,OAAdG,OAAcH,OAARI,OAAQJ;IAC7B,IAAMK,SAASD,SAASX,aAAaW,SAAS,MAAM,KAAKA,KAAKT,OAAO,CAAC,KAAK;IAC3E,IAAMW,SAASD,WAAW,KAAMD,iBAAAA,kBAAAA,OAAQ,KAAMC,OAAOE,MAAM,KAAK,IAAI,AAAC,GAAS,OAAPF,QAAO,SAAO,AAAC,GAAwBA,OAAtBA,OAAOG,KAAK,CAAC,GAAG,IAAG,KAAmB,OAAhBH,OAAOG,KAAK,CAAC;IAC3H,IAAMC,aAAa,AAAC,GAAUN,OAARD,MAAK,KAAUI,OAAPH,MAAc,OAAPG;IACrC,OAAOI,OAAOC,KAAK,CAACC,KAAKC,KAAK,CAACJ,eAAejB,QAAQiB;AACxD;AAIO,SAASxB,SAASO,KAAc;IACrC,IAAIA,UAAU,QAAQA,UAAUC,WAAW,OAAO;IAClD,IAAI,OAAOD,UAAU,WAAW,OAAOsB,OAAOtB,QAAQ,IAAI;IAC1D,IAAI,OAAOA,UAAU,UAAU,OAAOkB,OAAOK,aAAa,CAACvB,SAASsB,OAAOtB,SAASA;IACpF,IAAI,OAAOA,UAAU,UAAU,OAAON,cAAcM;IACpD,OAAOwB,KAAKC,SAAS,CAACzB;AACxB;AAIO,SAAST,WAAWmC,CAAS;IAClC,IAAIC,IAAI;IACR,IAAK,IAAIC,IAAI,GAAGA,IAAIF,EAAEX,MAAM,EAAEa,IAAK,IAAIF,EAAEG,UAAU,CAACD,OAAO,IAAID;IAC/D,OAAOA;AACT;AAGO,SAAS9B,iBAAiBiC,GAAW;IAC1C,IAAMC,OAAOD,IAAIE,KAAK,CAAC;IACvB,IAAI,CAACD,MAAM,OAAO;QAAEE,IAAI;QAAMC,MAAMJ;IAAI;IACxC,IAAMK,OAAOL,IAAId,KAAK,CAACe,IAAI,CAAC,EAAE,CAAChB,MAAM;IACrC,IAAMqB,QAAQD,KAAKH,KAAK,CAAC;IACzB,IAAI,CAACI,SAASA,MAAMC,KAAK,KAAKpC,WAAW,OAAO;QAAEgC,IAAI;QAAMC,MAAMJ;IAAI;IACtE,OAAO;QAAEG,IAAIE,KAAKnB,KAAK,CAAC,GAAGoB,MAAMC,KAAK;QAAGH,MAAMC,KAAKnB,KAAK,CAACoB,MAAMC,KAAK,GAAGD,KAAK,CAAC,EAAE,CAACrB,MAAM;IAAE;AAC3F;AAEA,qLAAqL;AACrL,2IAA2I;AAC3I,SAASuB,oBAAoBC,OAAe,EAAEC,GAAqC,EAAEC,QAAkB;IACrG,IAAIC,QAAQ;IACZ,4FAA4F;IAC5F,yCAAyC;IACzCC,IAAAA,WAAK,EAACH,KAAK;QACTI,MAAAA,SAAAA,KAAKC,IAAI,EAAEC,IAAI;YACb,IAAI,CAACC,IAAAA,kBAAY,EAACD,KAAKE,GAAG,GAAG,OAAO/C;YACpCyC,QAAQ;YACR,OAAOC,WAAK,CAACM,KAAK;QACpB;IACF;IACA,8EAA8E;IAC9E,IAAIP,OAAOD,SAASS,IAAI,CAAC,AAAC,YAAmB,OAARX,SAAQ;AAC/C;AAEA,oSAAoS;AACpS,6IAA6I;AAC7I,SAASY,UAAUC,OAAe;IAChC,OAAOA,QAAQC,KAAK,CAAC,KAAK,CAAC,EAAE,CAAClD,OAAO,CAAC,SAAS;AACjD;AAEO,SAASP,iBAAiB2C,OAAe,EAAEN,EAAU,EAAEQ,QAAkB;IAC9E,yFAAyF;IACzF,wDAAwD;IACxD,IAAMD,MAAMc,IAAAA,mBAAa,EAACrB,IAAI;QAAEsB,UAAU;IAAS;IACnD,IAAMC,UAAUhB,IAAIiB,MAAM,CAACC,MAAM,CAAC,SAACC;eAAQ,CAAC5D,oBAAoB6D,GAAG,CAACD,IAAIE,IAAI;;IAC5E,IAAIL,QAAQzC,MAAM,GAAG,GAAG;QACtB,IAAM+C,SAASN,QAAQzC,MAAM,GAAG,IAAI,AAAC,SAA2B,OAAnByC,QAAQzC,MAAM,GAAG,GAAE,YAAU;QAC1E,IAAMgD,aAAa,AAAC,GAAkCD,OAAhCX,UAAUK,OAAO,CAAC,EAAE,CAACJ,OAAO,GAAW,OAAPU;QACtDrB,SAASS,IAAI,CAAC,AAAC,YAA2Ea,OAAhExB,SAAQ,0DAAmE,OAAXwB;QAC1F,OAAO;YAAEC,MAAM,CAAC;YAAGD,YAAAA;QAAW;IAChC;IAEA,IAAIC;IACJ,IAAI;QACFA,OAAOxB,IAAIyB,IAAI;IACjB,EAAE,OAAON,KAAK;QACZ,uFAAuF;QACvF,kEAAkE;QAClE,IAAMI,cAAaZ,UAAU,AAACQ,IAAcP,OAAO;QACnDX,SAASS,IAAI,CAAC,AAAC,YAA2Ea,OAAhExB,SAAQ,0DAAmE,OAAXwB;QAC1F,OAAO;YAAEC,MAAM,CAAC;YAAGD,YAAAA;QAAW;IAChC;IAEA,IAAIC,SAAS,QAAQA,SAAS/D,WAAW,OAAO;QAAE+D,MAAM,CAAC;QAAGD,YAAY;IAAK;IAC7E,IAAI,CAAA,OAAOC,qCAAP,SAAOA,KAAG,MAAM,YAAYE,MAAMC,OAAO,CAACH,OAAO;QACnD,IAAMD,cAAa;QACnBtB,SAASS,IAAI,CAAC,AAAC,YAAsBa,OAAXxB,SAAQ,KAAc,OAAXwB,aAAW;QAChD,OAAO;YAAEC,MAAM,CAAC;YAAGD,YAAAA;QAAW;IAChC;IACAzB,oBAAoBC,SAASC,KAAKC;IAClC,OAAO;QAAEuB,MAAMA;QAAiCD,YAAY;IAAK;AACnE"}
+{"version":3,"sources":["/Users/kevin/Dev/OpenSource/ai/sensemaking/src/scan/frontmatter.ts"],"sourcesContent":["import { isCollection, parseDocument, visit } from 'yaml';\n\n// Frontmatter keys that would collide with table columns. Exported so reconcile.ts's upsert can tell\n// a feature-owned column (`_rank`) from a parsed one and leave it alone on reparse.\nexport const RESERVED_COLUMNS = new Set(['path', '_mtime', '_ctime', '_size', '_rank', '_parse_error', 'content', 'links', 'sections']);\n\n// Only BAD_SCALAR_START qualifies: YAML 1.2 reserves `@` and `` ` `` at the start of a plain scalar, so `aliases: [@handle]` can only mean [\"@handle\"].\n// Every other code has a second reading (a swallowed key, a dropped URL, a silently overwritten duplicate) that would write values nobody wrote.\nconst ACCEPTED_YAML_CODES = new Set(['BAD_SCALAR_START']);\n\nexport function normalizeText(value: unknown): string {\n  if (value === null || value === undefined) return '';\n  return String(value).replace(/\\s+/g, ' ').trim();\n}\n\n// SQLite's datetime() rejects a colonless offset (`-0800`) and a space separator, which ISO 8601\n// allows and producers emit. A rejected date is invisible, not excluded: every comparison is NULL.\nconst ISO_DATETIME = /^(\\d{4}-\\d{2}-\\d{2})[T ](\\d{2}:\\d{2}(?::\\d{2})?(?:\\.\\d+)?)(Z|[+-]\\d{2}(?::?\\d{2})?)?$/;\n\n// A value opening `YYYY-MM-DDT` was meant to be a datetime; prose never is. Reported when it\n// cannot be normalized, so a typo surfaces on the next crawl instead of at some later audit.\nconst MEANT_AS_DATETIME = /^\\d{4}-\\d{2}-\\d{2}[T ]\\d/;\n\nexport function looksLikeDatetime(value: string): boolean {\n  return MEANT_AS_DATETIME.test(value);\n}\n\n// Punctuation only, never a timezone conversion: the offset survives, so substr(d,1,10) is still\n// the local date. A shape that is not a real instant is left as written, and stays auditable.\nexport function normalizeDate(value: string): string {\n  const m = ISO_DATETIME.exec(value);\n  if (m === null) return value;\n  const [, date, time, zone] = m;\n  const digits = zone === undefined || zone === 'Z' ? '' : zone.replace(':', '');\n  const offset = digits === '' ? (zone ?? '') : digits.length === 3 ? `${digits}:00` : `${digits.slice(0, 3)}:${digits.slice(3)}`;\n  const normalized = `${date}T${time}${offset}`;\n  return Number.isNaN(Date.parse(normalized)) ? value : normalized;\n}\n\n// Storage class follows the YAML scalar. Booleans store as 1/0, so `WHERE flag = 1` matches\n// and `WHERE flag = 'true'` cannot; `map` prints observed types so the mismatch is visible.\nexport function mapValue(value: unknown): string | number | bigint | null {\n  if (value === null || value === undefined) return null;\n  if (typeof value === 'boolean') return BigInt(value ? 1 : 0);\n  if (typeof value === 'number') return Number.isSafeInteger(value) ? BigInt(value) : value;\n  if (typeof value === 'string') return normalizeDate(value);\n  return JSON.stringify(value);\n}\n\n// Exactly `s.split('\\n').length` without allocating the array: callers that need only the count\n// run it per file per feature on the parse hot path.\nexport function countLines(s: string): number {\n  let n = 1;\n  for (let i = 0; i < s.length; i++) if (s.charCodeAt(i) === 10) n++;\n  return n;\n}\n\n// The delimiter split is all this package used gray-matter for.\nexport function splitFrontmatter(raw: string): { fm: string | null; body: string } {\n  const open = raw.match(/^---\\r?\\n/);\n  if (!open) return { fm: null, body: raw };\n  const rest = raw.slice(open[0].length);\n  const close = rest.match(/^---\\r?(\\n|$)/m);\n  if (!close || close.index === undefined) return { fm: null, body: raw };\n  return { fm: rest.slice(0, close.index), body: rest.slice(close.index + close[0].length) };\n}\n\n// A well-formed document can hold a value nobody meant: `created: {{date}}` is valid YAML for a flow map used as a mapping key, so it raises no error and stores {\"{ date }\": null}.\n// No error code catches that; yaml's own warning lacks a path and fires once per document, so this reports it here instead, with the path.\nfunction warnStringifiedKeys(relPath: string, doc: ReturnType<typeof parseDocument>, warnings: string[]): void {\n  let found = false;\n  // Nested, not top level: `created: {{date}}` puts the collection key one level down, inside\n  // the flow map that `{{...}}` parses as.\n  visit(doc, {\n    Pair(_key, pair) {\n      if (!isCollection(pair.key)) return undefined;\n      found = true;\n      return visit.BREAK;\n    },\n  });\n  // One per file: a template repeats the same mistake on every field it stamps.\n  if (found) warnings.push(`warning: ${relPath} frontmatter has a key that is itself a list or mapping, stored as text; this is usually an unrendered template placeholder like {{date}}`);\n}\n\n// Anything past ACCEPTED_YAML_CODES is quarantined: no frontmatter columns, `_parse_error` carries the reason. Recovering it would write values nobody wrote, worse than absence since no query can see it; content, links and sections never touch frontmatter, so a broken note stays searchable.\n// yaml's message continues onto a source snippet, so the first line is the sentence, minus the colon that introduced the part being dropped.\nfunction firstLine(message: string): string {\n  return message.split('\\n')[0].replace(/:\\s*$/, '');\n}\n\nexport function parseFrontmatter(relPath: string, fm: string, warnings: string[]): { data: Record<string, unknown>; parseError: string | null } {\n  // logLevel silences yaml's own pathless warnings; warnStringifiedKeys re-reports the one\n  // that carries information, with the file it came from.\n  const doc = parseDocument(fm, { logLevel: 'silent' });\n  const refused = doc.errors.filter((err) => !ACCEPTED_YAML_CODES.has(err.code));\n  if (refused.length > 0) {\n    const detail = refused.length > 1 ? ` (and ${refused.length - 1} more)` : '';\n    const parseError = `${firstLine(refused[0].message)}${detail}`;\n    warnings.push(`warning: ${relPath} frontmatter did not parse, so none of it is indexed: ${parseError}`);\n    return { data: {}, parseError };\n  }\n\n  let data: unknown;\n  try {\n    data = doc.toJS();\n  } catch (err) {\n    // Reaches here with doc.errors empty: `title: **Bold**` parses, then opens an alias on\n    // materialisation. An empty error list is not a successful parse.\n    const parseError = firstLine((err as Error).message);\n    warnings.push(`warning: ${relPath} frontmatter did not parse, so none of it is indexed: ${parseError}`);\n    return { data: {}, parseError };\n  }\n\n  if (data === null || data === undefined) return { data: {}, parseError: null };\n  if (typeof data !== 'object' || Array.isArray(data)) {\n    const parseError = 'frontmatter is not a key-value mapping';\n    warnings.push(`warning: ${relPath} ${parseError}; none of it is indexed`);\n    return { data: {}, parseError };\n  }\n  warnStringifiedKeys(relPath, doc, warnings);\n  return { data: data as Record<string, unknown>, parseError: null };\n}\n"],"names":["RESERVED_COLUMNS","countLines","looksLikeDatetime","mapValue","normalizeDate","normalizeText","parseFrontmatter","splitFrontmatter","Set","ACCEPTED_YAML_CODES","value","undefined","String","replace","trim","ISO_DATETIME","MEANT_AS_DATETIME","test","m","exec","date","time","zone","digits","offset","length","slice","normalized","Number","isNaN","Date","parse","BigInt","isSafeInteger","JSON","stringify","s","n","i","charCodeAt","raw","open","match","fm","body","rest","close","index","warnStringifiedKeys","relPath","doc","warnings","found","visit","Pair","_key","pair","isCollection","key","BREAK","push","firstLine","message","split","parseDocument","logLevel","refused","errors","filter","err","has","code","detail","parseError","data","toJS","Array","isArray"],"mappings":";;;;;;;;;;;QAIaA;eAAAA;;QA+CGC;eAAAA;;QA5BAC;eAAAA;;QAkBAC;eAAAA;;QAZAC;eAAAA;;QAnBAC;eAAAA;;QAgFAC;eAAAA;;QAhCAC;eAAAA;;;oBA1DmC;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;AAI5C,IAAMP,mBAAmB,IAAIQ,IAAI;IAAC;IAAQ;IAAU;IAAU;IAAS;IAAS;IAAgB;IAAW;IAAS;CAAW;AAEtI,wJAAwJ;AACxJ,iJAAiJ;AACjJ,IAAMC,sBAAsB,IAAID,IAAI;IAAC;CAAmB;AAEjD,SAASH,cAAcK,KAAc;IAC1C,IAAIA,UAAU,QAAQA,UAAUC,WAAW,OAAO;IAClD,OAAOC,OAAOF,OAAOG,OAAO,CAAC,QAAQ,KAAKC,IAAI;AAChD;AAEA,iGAAiG;AACjG,mGAAmG;AACnG,IAAMC,eAAe;AAErB,6FAA6F;AAC7F,6FAA6F;AAC7F,IAAMC,oBAAoB;AAEnB,SAASd,kBAAkBQ,KAAa;IAC7C,OAAOM,kBAAkBC,IAAI,CAACP;AAChC;AAIO,SAASN,cAAcM,KAAa;IACzC,IAAMQ,IAAIH,aAAaI,IAAI,CAACT;IAC5B,IAAIQ,MAAM,MAAM,OAAOR;IACvB,IAA6BQ,sBAAAA,OAApBE,OAAoBF,OAAdG,OAAcH,OAARI,OAAQJ;IAC7B,IAAMK,SAASD,SAASX,aAAaW,SAAS,MAAM,KAAKA,KAAKT,OAAO,CAAC,KAAK;IAC3E,IAAMW,SAASD,WAAW,KAAMD,iBAAAA,kBAAAA,OAAQ,KAAMC,OAAOE,MAAM,KAAK,IAAI,AAAC,GAAS,OAAPF,QAAO,SAAO,AAAC,GAAwBA,OAAtBA,OAAOG,KAAK,CAAC,GAAG,IAAG,KAAmB,OAAhBH,OAAOG,KAAK,CAAC;IAC3H,IAAMC,aAAa,AAAC,GAAUN,OAARD,MAAK,KAAUI,OAAPH,MAAc,OAAPG;IACrC,OAAOI,OAAOC,KAAK,CAACC,KAAKC,KAAK,CAACJ,eAAejB,QAAQiB;AACxD;AAIO,SAASxB,SAASO,KAAc;IACrC,IAAIA,UAAU,QAAQA,UAAUC,WAAW,OAAO;IAClD,IAAI,OAAOD,UAAU,WAAW,OAAOsB,OAAOtB,QAAQ,IAAI;IAC1D,IAAI,OAAOA,UAAU,UAAU,OAAOkB,OAAOK,aAAa,CAACvB,SAASsB,OAAOtB,SAASA;IACpF,IAAI,OAAOA,UAAU,UAAU,OAAON,cAAcM;IACpD,OAAOwB,KAAKC,SAAS,CAACzB;AACxB;AAIO,SAAST,WAAWmC,CAAS;IAClC,IAAIC,IAAI;IACR,IAAK,IAAIC,IAAI,GAAGA,IAAIF,EAAEX,MAAM,EAAEa,IAAK,IAAIF,EAAEG,UAAU,CAACD,OAAO,IAAID;IAC/D,OAAOA;AACT;AAGO,SAAS9B,iBAAiBiC,GAAW;IAC1C,IAAMC,OAAOD,IAAIE,KAAK,CAAC;IACvB,IAAI,CAACD,MAAM,OAAO;QAAEE,IAAI;QAAMC,MAAMJ;IAAI;IACxC,IAAMK,OAAOL,IAAId,KAAK,CAACe,IAAI,CAAC,EAAE,CAAChB,MAAM;IACrC,IAAMqB,QAAQD,KAAKH,KAAK,CAAC;IACzB,IAAI,CAACI,SAASA,MAAMC,KAAK,KAAKpC,WAAW,OAAO;QAAEgC,IAAI;QAAMC,MAAMJ;IAAI;IACtE,OAAO;QAAEG,IAAIE,KAAKnB,KAAK,CAAC,GAAGoB,MAAMC,KAAK;QAAGH,MAAMC,KAAKnB,KAAK,CAACoB,MAAMC,KAAK,GAAGD,KAAK,CAAC,EAAE,CAACrB,MAAM;IAAE;AAC3F;AAEA,qLAAqL;AACrL,2IAA2I;AAC3I,SAASuB,oBAAoBC,OAAe,EAAEC,GAAqC,EAAEC,QAAkB;IACrG,IAAIC,QAAQ;IACZ,4FAA4F;IAC5F,yCAAyC;IACzCC,IAAAA,WAAK,EAACH,KAAK;QACTI,MAAAA,SAAAA,KAAKC,IAAI,EAAEC,IAAI;YACb,IAAI,CAACC,IAAAA,kBAAY,EAACD,KAAKE,GAAG,GAAG,OAAO/C;YACpCyC,QAAQ;YACR,OAAOC,WAAK,CAACM,KAAK;QACpB;IACF;IACA,8EAA8E;IAC9E,IAAIP,OAAOD,SAASS,IAAI,CAAC,AAAC,YAAmB,OAARX,SAAQ;AAC/C;AAEA,oSAAoS;AACpS,6IAA6I;AAC7I,SAASY,UAAUC,OAAe;IAChC,OAAOA,QAAQC,KAAK,CAAC,KAAK,CAAC,EAAE,CAAClD,OAAO,CAAC,SAAS;AACjD;AAEO,SAASP,iBAAiB2C,OAAe,EAAEN,EAAU,EAAEQ,QAAkB;IAC9E,yFAAyF;IACzF,wDAAwD;IACxD,IAAMD,MAAMc,IAAAA,mBAAa,EAACrB,IAAI;QAAEsB,UAAU;IAAS;IACnD,IAAMC,UAAUhB,IAAIiB,MAAM,CAACC,MAAM,CAAC,SAACC;eAAQ,CAAC5D,oBAAoB6D,GAAG,CAACD,IAAIE,IAAI;;IAC5E,IAAIL,QAAQzC,MAAM,GAAG,GAAG;QACtB,IAAM+C,SAASN,QAAQzC,MAAM,GAAG,IAAI,AAAC,SAA2B,OAAnByC,QAAQzC,MAAM,GAAG,GAAE,YAAU;QAC1E,IAAMgD,aAAa,AAAC,GAAkCD,OAAhCX,UAAUK,OAAO,CAAC,EAAE,CAACJ,OAAO,GAAW,OAAPU;QACtDrB,SAASS,IAAI,CAAC,AAAC,YAA2Ea,OAAhExB,SAAQ,0DAAmE,OAAXwB;QAC1F,OAAO;YAAEC,MAAM,CAAC;YAAGD,YAAAA;QAAW;IAChC;IAEA,IAAIC;IACJ,IAAI;QACFA,OAAOxB,IAAIyB,IAAI;IACjB,EAAE,OAAON,KAAK;QACZ,uFAAuF;QACvF,kEAAkE;QAClE,IAAMI,cAAaZ,UAAU,AAACQ,IAAcP,OAAO;QACnDX,SAASS,IAAI,CAAC,AAAC,YAA2Ea,OAAhExB,SAAQ,0DAAmE,OAAXwB;QAC1F,OAAO;YAAEC,MAAM,CAAC;YAAGD,YAAAA;QAAW;IAChC;IAEA,IAAIC,SAAS,QAAQA,SAAS/D,WAAW,OAAO;QAAE+D,MAAM,CAAC;QAAGD,YAAY;IAAK;IAC7E,IAAI,CAAA,OAAOC,qCAAP,SAAOA,KAAG,MAAM,YAAYE,MAAMC,OAAO,CAACH,OAAO;QACnD,IAAMD,cAAa;QACnBtB,SAASS,IAAI,CAAC,AAAC,YAAsBa,OAAXxB,SAAQ,KAAc,OAAXwB,aAAW;QAChD,OAAO;YAAEC,MAAM,CAAC;YAAGD,YAAAA;QAAW;IAChC;IACAzB,oBAAoBC,SAASC,KAAKC;IAClC,OAAO;QAAEuB,MAAMA;QAAiCD,YAAY;IAAK;AACnE"}

package/dist/cjs/scan/index.js

@@ -35,6 +35,7 @@ var _frontmatterts = require("./frontmatter.js");
 var _listts = require("./list.js");
 function parseFile(file) {
     var extractors = arguments.length > 1 && arguments[1] !== void 0 ? arguments[1] : [], cfg = arguments.length > 2 ? arguments[2] : void 0;
+    // Preserve the authored bytes: feature extractors and section offsets use the raw document.
     var raw = (0, _nodefs.readFileSync)(file.absPath, 'utf8');
     var warnings = [];
     var _splitFrontmatter = (0, _frontmatterts.splitFrontmatter)(raw), fm = _splitFrontmatter.fm, content = _splitFrontmatter.body;

package/dist/cjs/scan/index.js.map

@@ -1 +1 @@
-{"version":3,"sources":["/Users/kevin/Dev/OpenSource/ai/sensemaking/src/scan/index.ts"],"sourcesContent":["import { readFileSync } from 'node:fs';\nimport { parse } from '../chunk/index.ts';\nimport type { Config } from '../config/index.ts';\nimport type { Feature } from '../features/types.ts';\nimport { textFromBlocks } from '../text/strip.ts';\nimport { looksLikeDatetime, mapValue, normalizeText, parseFrontmatter, RESERVED_COLUMNS, splitFrontmatter } from './frontmatter.ts';\nimport type { FileStat } from './list.ts';\n\n// Filesystem -> rows. Pure data in, data + warnings out; store/sqlite/reconcile.ts does the SQL.\n\nexport { looksLikeDatetime, normalizeDate, RESERVED_COLUMNS } from './frontmatter.ts';\nexport type { FileStat } from './list.ts';\nexport { listFiles, toPosixPath } from './list.ts';\n\nexport interface ParsedDoc {\n  relPath: string;\n  mtimeMs: number;\n  ctimeMs: number;\n  size: number;\n  presets: string[];\n  data: Record<string, string | number | bigint | null>;\n  // NULL when the frontmatter parsed, the first YAML message otherwise. In the row rather than\n  // a side table so `SELECT *` and any `IS NULL` investigation trip over it without being asked.\n  parseError: string | null;\n  // title/summary are duplicated from frontmatter so bm25() can weight them above the body text.\n  search: { title: string; summary: string; text: string };\n  // Per-feature extraction results, keyed by feature name; features store them at reconcile.\n  extracted: Record<string, unknown>;\n}\n\nexport function parseFile(file: FileStat, extractors: Feature[] = [], cfg?: Config): { doc: ParsedDoc; warnings: string[] } {\n  const raw = readFileSync(file.absPath, 'utf8');\n  const warnings: string[] = [];\n\n  const { fm, body: content } = splitFrontmatter(raw);\n  const { data, parseError } = fm === null ? { data: {} as Record<string, unknown>, parseError: null } : parseFrontmatter(file.relPath, fm, warnings);\n  const mapped: Record<string, string | number | bigint | null> = {};\n\n  for (const key of Object.keys(data)) {\n    if (RESERVED_COLUMNS.has(key)) {\n      warnings.push(`warning: ${file.relPath} has a frontmatter key named \"${key}\", which is reserved; ignoring it`);\n      continue;\n    }\n    const value = mapValue(data[key]);\n    if (typeof value === 'string' && looksLikeDatetime(value) && Number.isNaN(Date.parse(value))) {\n      warnings.push(`warning: ${file.relPath}: ${key} is not a valid date (${value}), so it is invisible to every date comparison`);\n    }\n    mapped[key] = value;\n  }\n\n  // Parsed once and shared: the FTS text path (textFromBlocks, ~ strip.ts) and any feature\n  // that needs a parse tree (embed's chunker) both read this same block list.\n  const blocks = parse(content);\n\n  // title/summary are plain YAML strings -- whitespace-collapse only;\n  // the prose gets the full markdown strip.\n  const search = { title: normalizeText(data.title), summary: normalizeText(data.summary), text: textFromBlocks(blocks) };\n\n  return {\n    doc: {\n      relPath: file.relPath,\n      mtimeMs: file.mtimeMs,\n      ctimeMs: file.ctimeMs,\n      size: file.size,\n      presets: file.presets,\n      data: mapped,\n      parseError,\n      search,\n      extracted: Object.fromEntries(extractors.filter((f) => f.extract).map((f) => [f.name, f.extract?.(raw, content, search, data, cfg, blocks)])),\n    },\n    warnings,\n  };\n}\n"],"names":["RESERVED_COLUMNS","listFiles","looksLikeDatetime","normalizeDate","parseFile","toPosixPath","file","extractors","cfg","raw","readFileSync","absPath","warnings","splitFrontmatter","fm","body","content","data","parseError","parseFrontmatter","relPath","mapped","Object","keys","key","has","push","value","mapValue","Number","isNaN","Date","parse","blocks","search","title","normalizeText","summary","text","textFromBlocks","doc","mtimeMs","ctimeMs","size","presets","extracted","fromEntries","filter","f","extract","map","name"],"mappings":";;;;;;;;;;;QAU2CA;eAAAA,+BAAgB;;QAElDC;eAAAA,iBAAS;;QAFTC;eAAAA,gCAAiB;;QAAEC;eAAAA,4BAAa;;QAoBzBC;eAAAA;;QAlBIC;eAAAA,mBAAW;;;sBAZF;uBACP;uBAGS;6BACkF;sBAO1E;AAkBhC,SAASD,UAAUE,IAAc;QAAEC,aAAAA,iEAAwB,EAAE,EAAEC;IACpE,IAAMC,MAAMC,IAAAA,oBAAY,EAACJ,KAAKK,OAAO,EAAE;IACvC,IAAMC,WAAqB,EAAE;IAE7B,IAA8BC,oBAAAA,IAAAA,+BAAgB,EAACJ,MAAvCK,KAAsBD,kBAAtBC,IAAIC,AAAMC,UAAYH,kBAAlBE;IACZ,IAA6BD,OAAAA,OAAO,OAAO;QAAEG,MAAM,CAAC;QAA8BC,YAAY;IAAK,IAAIC,IAAAA,+BAAgB,EAACb,KAAKc,OAAO,EAAEN,IAAIF,WAAlIK,OAAqBH,KAArBG,MAAMC,aAAeJ,KAAfI;IACd,IAAMG,SAA0D,CAAC;QAE5D,kCAAA,2BAAA;;QAAL,QAAK,YAAaC,OAAOC,IAAI,CAACN,0BAAzB,SAAA,6BAAA,QAAA,yBAAA,iCAAgC;YAAhC,IAAMO,MAAN;YACH,IAAIxB,+BAAgB,CAACyB,GAAG,CAACD,MAAM;gBAC7BZ,SAASc,IAAI,CAAC,AAAC,YAAwDF,OAA7ClB,KAAKc,OAAO,EAAC,kCAAoC,OAAJI,KAAI;gBAC3E;YACF;YACA,IAAMG,QAAQC,IAAAA,uBAAQ,EAACX,IAAI,CAACO,IAAI;YAChC,IAAI,OAAOG,UAAU,YAAYzB,IAAAA,gCAAiB,EAACyB,UAAUE,OAAOC,KAAK,CAACC,KAAKC,KAAK,CAACL,SAAS;gBAC5Ff,SAASc,IAAI,CAAC,AAAC,YAA4BF,OAAjBlB,KAAKc,OAAO,EAAC,MAAgCO,OAA5BH,KAAI,0BAA8B,OAANG,OAAM;YAC/E;YACAN,MAAM,CAACG,IAAI,GAAGG;QAChB;;QAVK;QAAA;;;iBAAA,6BAAA;gBAAA;;;gBAAA;sBAAA;;;;IAYL,yFAAyF;IACzF,4EAA4E;IAC5E,IAAMM,SAASD,IAAAA,cAAK,EAAChB;IAErB,oEAAoE;IACpE,0CAA0C;IAC1C,IAAMkB,SAAS;QAAEC,OAAOC,IAAAA,4BAAa,EAACnB,KAAKkB,KAAK;QAAGE,SAASD,IAAAA,4BAAa,EAACnB,KAAKoB,OAAO;QAAGC,MAAMC,IAAAA,uBAAc,EAACN;IAAQ;IAEtH,OAAO;QACLO,KAAK;YACHpB,SAASd,KAAKc,OAAO;YACrBqB,SAASnC,KAAKmC,OAAO;YACrBC,SAASpC,KAAKoC,OAAO;YACrBC,MAAMrC,KAAKqC,IAAI;YACfC,SAAStC,KAAKsC,OAAO;YACrB3B,MAAMI;YACNH,YAAAA;YACAgB,QAAAA;YACAW,WAAWvB,OAAOwB,WAAW,CAACvC,WAAWwC,MAAM,CAAC,SAACC;uBAAMA,EAAEC,OAAO;eAAEC,GAAG,CAAC,SAACF;oBAAeA;uBAAT;oBAACA,EAAEG,IAAI;qBAAEH,aAAAA,EAAEC,OAAO,cAATD,iCAAAA,gBAAAA,GAAYvC,KAAKO,SAASkB,QAAQjB,MAAMT,KAAKyB;iBAAQ;;QAC7I;QACArB,UAAAA;IACF;AACF"}
+{"version":3,"sources":["/Users/kevin/Dev/OpenSource/ai/sensemaking/src/scan/index.ts"],"sourcesContent":["import { readFileSync } from 'node:fs';\nimport { parse } from '../chunk/index.ts';\nimport type { Config } from '../config/index.ts';\nimport type { Feature } from '../features/types.ts';\nimport { textFromBlocks } from '../text/strip.ts';\nimport { looksLikeDatetime, mapValue, normalizeText, parseFrontmatter, RESERVED_COLUMNS, splitFrontmatter } from './frontmatter.ts';\nimport type { FileStat } from './list.ts';\n\n// Filesystem -> rows. Pure data in, data + warnings out; store/sqlite/reconcile.ts does the SQL.\n\nexport { looksLikeDatetime, normalizeDate, RESERVED_COLUMNS } from './frontmatter.ts';\nexport type { FileStat } from './list.ts';\nexport { listFiles, toPosixPath } from './list.ts';\n\nexport interface ParsedDoc {\n  relPath: string;\n  mtimeMs: number;\n  ctimeMs: number;\n  size: number;\n  presets: string[];\n  data: Record<string, string | number | bigint | null>;\n  // NULL when the frontmatter parsed, the first YAML message otherwise. In the row rather than\n  // a side table so `SELECT *` and any `IS NULL` investigation trip over it without being asked.\n  parseError: string | null;\n  // title/summary are duplicated from frontmatter so bm25() can weight them above the body text.\n  search: { title: string; summary: string; text: string };\n  // Per-feature extraction results, keyed by feature name; features store them at reconcile.\n  extracted: Record<string, unknown>;\n}\n\nexport function parseFile(file: FileStat, extractors: Feature[] = [], cfg?: Config): { doc: ParsedDoc; warnings: string[] } {\n  // Preserve the authored bytes: feature extractors and section offsets use the raw document.\n  const raw = readFileSync(file.absPath, 'utf8');\n  const warnings: string[] = [];\n\n  const { fm, body: content } = splitFrontmatter(raw);\n  const { data, parseError } = fm === null ? { data: {} as Record<string, unknown>, parseError: null } : parseFrontmatter(file.relPath, fm, warnings);\n  const mapped: Record<string, string | number | bigint | null> = {};\n\n  for (const key of Object.keys(data)) {\n    if (RESERVED_COLUMNS.has(key)) {\n      warnings.push(`warning: ${file.relPath} has a frontmatter key named \"${key}\", which is reserved; ignoring it`);\n      continue;\n    }\n    const value = mapValue(data[key]);\n    if (typeof value === 'string' && looksLikeDatetime(value) && Number.isNaN(Date.parse(value))) {\n      warnings.push(`warning: ${file.relPath}: ${key} is not a valid date (${value}), so it is invisible to every date comparison`);\n    }\n    mapped[key] = value;\n  }\n\n  // Parsed once and shared: the FTS text path (textFromBlocks, ~ strip.ts) and any feature\n  // that needs a parse tree (embed's chunker) both read this same block list.\n  const blocks = parse(content);\n\n  // title/summary are plain YAML strings -- whitespace-collapse only;\n  // the prose gets the full markdown strip.\n  const search = { title: normalizeText(data.title), summary: normalizeText(data.summary), text: textFromBlocks(blocks) };\n\n  return {\n    doc: {\n      relPath: file.relPath,\n      mtimeMs: file.mtimeMs,\n      ctimeMs: file.ctimeMs,\n      size: file.size,\n      presets: file.presets,\n      data: mapped,\n      parseError,\n      search,\n      extracted: Object.fromEntries(extractors.filter((f) => f.extract).map((f) => [f.name, f.extract?.(raw, content, search, data, cfg, blocks)])),\n    },\n    warnings,\n  };\n}\n"],"names":["RESERVED_COLUMNS","listFiles","looksLikeDatetime","normalizeDate","parseFile","toPosixPath","file","extractors","cfg","raw","readFileSync","absPath","warnings","splitFrontmatter","fm","body","content","data","parseError","parseFrontmatter","relPath","mapped","Object","keys","key","has","push","value","mapValue","Number","isNaN","Date","parse","blocks","search","title","normalizeText","summary","text","textFromBlocks","doc","mtimeMs","ctimeMs","size","presets","extracted","fromEntries","filter","f","extract","map","name"],"mappings":";;;;;;;;;;;QAU2CA;eAAAA,+BAAgB;;QAElDC;eAAAA,iBAAS;;QAFTC;eAAAA,gCAAiB;;QAAEC;eAAAA,4BAAa;;QAoBzBC;eAAAA;;QAlBIC;eAAAA,mBAAW;;;sBAZF;uBACP;uBAGS;6BACkF;sBAO1E;AAkBhC,SAASD,UAAUE,IAAc;QAAEC,aAAAA,iEAAwB,EAAE,EAAEC;IACpE,4FAA4F;IAC5F,IAAMC,MAAMC,IAAAA,oBAAY,EAACJ,KAAKK,OAAO,EAAE;IACvC,IAAMC,WAAqB,EAAE;IAE7B,IAA8BC,oBAAAA,IAAAA,+BAAgB,EAACJ,MAAvCK,KAAsBD,kBAAtBC,IAAIC,AAAMC,UAAYH,kBAAlBE;IACZ,IAA6BD,OAAAA,OAAO,OAAO;QAAEG,MAAM,CAAC;QAA8BC,YAAY;IAAK,IAAIC,IAAAA,+BAAgB,EAACb,KAAKc,OAAO,EAAEN,IAAIF,WAAlIK,OAAqBH,KAArBG,MAAMC,aAAeJ,KAAfI;IACd,IAAMG,SAA0D,CAAC;QAE5D,kCAAA,2BAAA;;QAAL,QAAK,YAAaC,OAAOC,IAAI,CAACN,0BAAzB,SAAA,6BAAA,QAAA,yBAAA,iCAAgC;YAAhC,IAAMO,MAAN;YACH,IAAIxB,+BAAgB,CAACyB,GAAG,CAACD,MAAM;gBAC7BZ,SAASc,IAAI,CAAC,AAAC,YAAwDF,OAA7ClB,KAAKc,OAAO,EAAC,kCAAoC,OAAJI,KAAI;gBAC3E;YACF;YACA,IAAMG,QAAQC,IAAAA,uBAAQ,EAACX,IAAI,CAACO,IAAI;YAChC,IAAI,OAAOG,UAAU,YAAYzB,IAAAA,gCAAiB,EAACyB,UAAUE,OAAOC,KAAK,CAACC,KAAKC,KAAK,CAACL,SAAS;gBAC5Ff,SAASc,IAAI,CAAC,AAAC,YAA4BF,OAAjBlB,KAAKc,OAAO,EAAC,MAAgCO,OAA5BH,KAAI,0BAA8B,OAANG,OAAM;YAC/E;YACAN,MAAM,CAACG,IAAI,GAAGG;QAChB;;QAVK;QAAA;;;iBAAA,6BAAA;gBAAA;;;gBAAA;sBAAA;;;;IAYL,yFAAyF;IACzF,4EAA4E;IAC5E,IAAMM,SAASD,IAAAA,cAAK,EAAChB;IAErB,oEAAoE;IACpE,0CAA0C;IAC1C,IAAMkB,SAAS;QAAEC,OAAOC,IAAAA,4BAAa,EAACnB,KAAKkB,KAAK;QAAGE,SAASD,IAAAA,4BAAa,EAACnB,KAAKoB,OAAO;QAAGC,MAAMC,IAAAA,uBAAc,EAACN;IAAQ;IAEtH,OAAO;QACLO,KAAK;YACHpB,SAASd,KAAKc,OAAO;YACrBqB,SAASnC,KAAKmC,OAAO;YACrBC,SAASpC,KAAKoC,OAAO;YACrBC,MAAMrC,KAAKqC,IAAI;YACfC,SAAStC,KAAKsC,OAAO;YACrB3B,MAAMI;YACNH,YAAAA;YACAgB,QAAAA;YACAW,WAAWvB,OAAOwB,WAAW,CAACvC,WAAWwC,MAAM,CAAC,SAACC;uBAAMA,EAAEC,OAAO;eAAEC,GAAG,CAAC,SAACF;oBAAeA;uBAAT;oBAACA,EAAEG,IAAI;qBAAEH,aAAAA,EAAEC,OAAO,cAATD,iCAAAA,gBAAAA,GAAYvC,KAAKO,SAASkB,QAAQjB,MAAMT,KAAKyB;iBAAQ;;QAC7I;QACArB,UAAAA;IACF;AACF"}

package/dist/cjs/store/batch.d.cts

@@ -0,0 +1,4 @@
+export declare function rewriteInsert(sql: string, rowCount: number): {
+    sql: string;
+    width: number;
+} | null;

package/dist/cjs/store/batch.d.ts

@@ -0,0 +1,4 @@
+export declare function rewriteInsert(sql: string, rowCount: number): {
+    sql: string;
+    width: number;
+} | null;

package/dist/cjs/store/batch.js

@@ -0,0 +1,80 @@
+"use strict";
+Object.defineProperty(exports, "__esModule", {
+    value: true
+});
+Object.defineProperty(exports, "rewriteInsert", {
+    enumerable: true,
+    get: function() {
+        return rewriteInsert;
+    }
+});
+function _array_like_to_array(arr, len) {
+    if (len == null || len > arr.length) len = arr.length;
+    for(var i = 0, arr2 = new Array(len); i < len; i++)arr2[i] = arr[i];
+    return arr2;
+}
+function _array_with_holes(arr) {
+    if (Array.isArray(arr)) return arr;
+}
+function _iterable_to_array_limit(arr, i) {
+    var _i = arr == null ? null : typeof Symbol !== "undefined" && arr[Symbol.iterator] || arr["@@iterator"];
+    if (_i == null) return;
+    var _arr = [];
+    var _n = true;
+    var _d = false;
+    var _s, _e;
+    try {
+        for(_i = _i.call(arr); !(_n = (_s = _i.next()).done); _n = true){
+            _arr.push(_s.value);
+            if (i && _arr.length === i) break;
+        }
+    } catch (err) {
+        _d = true;
+        _e = err;
+    } finally{
+        try {
+            if (!_n && _i["return"] != null) _i["return"]();
+        } finally{
+            if (_d) throw _e;
+        }
+    }
+    return _arr;
+}
+function _non_iterable_rest() {
+    throw new TypeError("Invalid attempt to destructure non-iterable instance.\nIn order to be iterable, non-array objects must have a [Symbol.iterator]() method.");
+}
+function _sliced_to_array(arr, i) {
+    return _array_with_holes(arr) || _iterable_to_array_limit(arr, i) || _unsupported_iterable_to_array(arr, i) || _non_iterable_rest();
+}
+function _unsupported_iterable_to_array(o, minLen) {
+    if (!o) return;
+    if (typeof o === "string") return _array_like_to_array(o, minLen);
+    var n = Object.prototype.toString.call(o).slice(8, -1);
+    if (n === "Object" && o.constructor) n = o.constructor.name;
+    if (n === "Map" || n === "Set") return Array.from(n);
+    if (n === "Arguments" || /^(?:Ui|I)nt(?:8|16|32)(?:Clamped)?Array$/.test(n)) return _array_like_to_array(o, minLen);
+}
+// Rewrites a single-row parameterized INSERT into one multi-row statement. Engine-neutral string
+// work, shared by duckdb (batch.ts's rewriteBatch) and turso (turso/connection.ts's runBatch).
+var INSERT_RE = RegExp("^(INSERT\\s+(?:OR\\s+\\w+\\s+)?INTO\\s+\\S+\\s*\\([^)]*\\)\\s*VALUES\\s*)\\(([^()]*)\\)(.*)$", "is");
+function placeholderCount(tuple) {
+    var _tuple_match;
+    return ((_tuple_match = tuple.match(/\?/g)) !== null && _tuple_match !== void 0 ? _tuple_match : []).length;
+}
+function rewriteInsert(sql, rowCount) {
+    var m = INSERT_RE.exec(sql.trim());
+    if (!m) return null;
+    var _m = _sliced_to_array(m, 4), head = _m[1], tuple = _m[2], tail = _m[3];
+    var width = placeholderCount(tuple);
+    if (width === 0) return null;
+    var rows = Array.from({
+        length: rowCount
+    }, function() {
+        return "(".concat(tuple, ")");
+    }).join(', ');
+    return {
+        sql: "".concat(head).concat(rows).concat(tail),
+        width: width
+    };
+}
+/* CJS INTEROP */ if (exports.__esModule && exports.default) { try { Object.defineProperty(exports.default, '__esModule', { value: true }); for (var key in exports) { exports.default[key] = exports[key]; } } catch (_) {}; module.exports = exports.default; }

package/dist/cjs/store/batch.js.map

@@ -0,0 +1 @@
+{"version":3,"sources":["/Users/kevin/Dev/OpenSource/ai/sensemaking/src/store/batch.ts"],"sourcesContent":["// Rewrites a single-row parameterized INSERT into one multi-row statement. Engine-neutral string\n// work, shared by duckdb (batch.ts's rewriteBatch) and turso (turso/connection.ts's runBatch).\n\nconst INSERT_RE = /^(INSERT\\s+(?:OR\\s+\\w+\\s+)?INTO\\s+\\S+\\s*\\([^)]*\\)\\s*VALUES\\s*)\\(([^()]*)\\)(.*)$/is;\n\nfunction placeholderCount(tuple: string): number {\n  return (tuple.match(/\\?/g) ?? []).length;\n}\n\nexport function rewriteInsert(sql: string, rowCount: number): { sql: string; width: number } | null {\n  const m = INSERT_RE.exec(sql.trim());\n  if (!m) return null;\n  const [, head, tuple, tail] = m;\n  const width = placeholderCount(tuple);\n  if (width === 0) return null;\n  const rows = Array.from({ length: rowCount }, () => `(${tuple})`).join(', ');\n  return { sql: `${head}${rows}${tail}`, width };\n}\n"],"names":["rewriteInsert","INSERT_RE","placeholderCount","tuple","match","length","sql","rowCount","m","exec","trim","head","tail","width","rows","Array","from","join"],"mappings":";;;;+BASgBA;;;eAAAA;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;AAThB,iGAAiG;AACjG,+FAA+F;AAE/F,IAAMC,YAAY;AAElB,SAASC,iBAAiBC,KAAa;QAC7BA;IAAR,OAAO,EAACA,eAAAA,MAAMC,KAAK,CAAC,oBAAZD,0BAAAA,eAAsB,EAAE,EAAEE,MAAM;AAC1C;AAEO,SAASL,cAAcM,GAAW,EAAEC,QAAgB;IACzD,IAAMC,IAAIP,UAAUQ,IAAI,CAACH,IAAII,IAAI;IACjC,IAAI,CAACF,GAAG,OAAO;IACf,IAA8BA,sBAAAA,OAArBG,OAAqBH,OAAfL,QAAeK,OAARI,OAAQJ;IAC9B,IAAMK,QAAQX,iBAAiBC;IAC/B,IAAIU,UAAU,GAAG,OAAO;IACxB,IAAMC,OAAOC,MAAMC,IAAI,CAAC;QAAEX,QAAQE;IAAS,GAAG;eAAM,AAAC,IAAS,OAANJ,OAAM;OAAIc,IAAI,CAAC;IACvE,OAAO;QAAEX,KAAK,AAAC,GAASQ,OAAPH,MAAcC,OAAPE,MAAY,OAALF;QAAQC,OAAAA;IAAM;AAC/C"}

package/dist/cjs/store/duckdb/batch.d.cts

@@ -1,7 +1,3 @@
-export declare function rewriteInsert(sql: string, rowCount: number): {
-    sql: string;
-    width: number;
-} | null;
 export declare function rewriteUpdate(sql: string, rowCount: number): {
     sql: string;
     width: number;

package/dist/cjs/store/duckdb/batch.d.ts

@@ -1,7 +1,3 @@
-export declare function rewriteInsert(sql: string, rowCount: number): {
-    sql: string;
-    width: number;
-} | null;
 export declare function rewriteUpdate(sql: string, rowCount: number): {
     sql: string;
     width: number;

package/dist/cjs/store/duckdb/batch.js

@@ -1,3 +1,6 @@
+// Rewrites the store's single-row parameterized INSERT/UPDATE/DELETE into one multi-row statement --
+// DuckDB's columnar engine pays for row-at-a-time writes, not just the round trip (PRINCIPLES: documented-means-tested).
+// rewriteInsert itself lives in the shared ../batch.ts: turso folds INSERTs the same way (3.36 B).
 "use strict";
 Object.defineProperty(exports, "__esModule", {
     value: true
@@ -15,13 +18,11 @@ _export(exports, {
     get rewriteDelete () {
         return rewriteDelete;
     },
-    get rewriteInsert () {
-        return rewriteInsert;
-    },
     get rewriteUpdate () {
         return rewriteUpdate;
     }
 });
+var _batchts = require("../batch.js");
 function _array_like_to_array(arr, len) {
     if (len == null || len > arr.length) len = arr.length;
     for(var i = 0, arr2 = new Array(len); i < len; i++)arr2[i] = arr[i];
@@ -82,15 +83,8 @@ function _unsupported_iterable_to_array(o, minLen) {
     if (n === "Map" || n === "Set") return Array.from(n);
     if (n === "Arguments" || /^(?:Ui|I)nt(?:8|16|32)(?:Clamped)?Array$/.test(n)) return _array_like_to_array(o, minLen);
 }
-// Rewrites the store's single-row parameterized INSERT/UPDATE/DELETE into one multi-row statement --
-// DuckDB's columnar engine pays for row-at-a-time writes, not just the round trip (PRINCIPLES: documented-means-tested).
-var INSERT_RE = RegExp("^(INSERT\\s+(?:OR\\s+\\w+\\s+)?INTO\\s+\\S+\\s*\\([^)]*\\)\\s*VALUES\\s*)\\(([^()]*)\\)(.*)$", "is");
 var UPDATE_RE = RegExp("^UPDATE\\s+(\\S+)\\s+SET\\s+(.+?)\\s+WHERE\\s+(.+)$", "is");
 var DELETE_RE = RegExp("^DELETE\\s+FROM\\s+(\\S+)\\s+WHERE\\s+(.+)$", "is");
-function placeholderCount(tuple) {
-    var _tuple_match;
-    return ((_tuple_match = tuple.match(/\?/g)) !== null && _tuple_match !== void 0 ? _tuple_match : []).length;
-}
 // "a = ?, b = ?" or "a = ? AND b = ?" -> ["a", "b"]. A clause that doesn't fit yields a shorter
 // list, which rewriteUpdate/rewriteDelete reject as a width mismatch.
 function equalityColumns(clause, sep) {
@@ -134,22 +128,6 @@ function tuples(rowCount, width) {
         return row;
     }).join(', ');
 }
-function rewriteInsert(sql, rowCount) {
-    var m = INSERT_RE.exec(sql.trim());
-    if (!m) return null;
-    var _m = _sliced_to_array(m, 4), head = _m[1], tuple = _m[2], tail = _m[3];
-    var width = placeholderCount(tuple);
-    if (width === 0) return null;
-    var rows = Array.from({
-        length: rowCount
-    }, function() {
-        return "(".concat(tuple, ")");
-    }).join(', ');
-    return {
-        sql: "".concat(head).concat(rows).concat(tail),
-        width: width
-    };
-}
 function rewriteUpdate(sql, rowCount) {
     var m = UPDATE_RE.exec(sql.trim());
     if (!m) return null;
@@ -192,6 +170,6 @@ function rewriteDelete(sql, rowCount) {
 }
 function rewriteBatch(sql, rowCount) {
     var _ref, _rewriteInsert;
-    return (_ref = (_rewriteInsert = rewriteInsert(sql, rowCount)) !== null && _rewriteInsert !== void 0 ? _rewriteInsert : rewriteUpdate(sql, rowCount)) !== null && _ref !== void 0 ? _ref : rewriteDelete(sql, rowCount);
+    return (_ref = (_rewriteInsert = (0, _batchts.rewriteInsert)(sql, rowCount)) !== null && _rewriteInsert !== void 0 ? _rewriteInsert : rewriteUpdate(sql, rowCount)) !== null && _ref !== void 0 ? _ref : rewriteDelete(sql, rowCount);
 }
 /* CJS INTEROP */ if (exports.__esModule && exports.default) { try { Object.defineProperty(exports.default, '__esModule', { value: true }); for (var key in exports) { exports.default[key] = exports[key]; } } catch (_) {}; module.exports = exports.default; }

package/dist/cjs/store/duckdb/batch.js.map

@@ -1 +1 @@
-{"version":3,"sources":["/Users/kevin/Dev/OpenSource/ai/sensemaking/src/store/duckdb/batch.ts"],"sourcesContent":["// Rewrites the store's single-row parameterized INSERT/UPDATE/DELETE into one multi-row statement --\n// DuckDB's columnar engine pays for row-at-a-time writes, not just the round trip (PRINCIPLES: documented-means-tested).\n\nconst INSERT_RE = /^(INSERT\\s+(?:OR\\s+\\w+\\s+)?INTO\\s+\\S+\\s*\\([^)]*\\)\\s*VALUES\\s*)\\(([^()]*)\\)(.*)$/is;\nconst UPDATE_RE = /^UPDATE\\s+(\\S+)\\s+SET\\s+(.+?)\\s+WHERE\\s+(.+)$/is;\nconst DELETE_RE = /^DELETE\\s+FROM\\s+(\\S+)\\s+WHERE\\s+(.+)$/is;\n\nfunction placeholderCount(tuple: string): number {\n  return (tuple.match(/\\?/g) ?? []).length;\n}\n\n// \"a = ?, b = ?\" or \"a = ? AND b = ?\" -> [\"a\", \"b\"]. A clause that doesn't fit yields a shorter\n// list, which rewriteUpdate/rewriteDelete reject as a width mismatch.\nfunction equalityColumns(clause: string, sep: RegExp): string[] | null {\n  const parts = clause.split(sep).map((p) => p.trim());\n  const cols: string[] = [];\n  for (const part of parts) {\n    const m = /^(.+?)=\\s*\\?$/.exec(part);\n    if (!m) return null;\n    cols.push(m[1].trim());\n  }\n  return cols;\n}\n\nfunction tuples(rowCount: number, width: number): string {\n  const row = `(${Array.from({ length: width }, () => '?').join(', ')})`;\n  return Array.from({ length: rowCount }, () => row).join(', ');\n}\n\nexport function rewriteInsert(sql: string, rowCount: number): { sql: string; width: number } | null {\n  const m = INSERT_RE.exec(sql.trim());\n  if (!m) return null;\n  const [, head, tuple, tail] = m;\n  const width = placeholderCount(tuple);\n  if (width === 0) return null;\n  const rows = Array.from({ length: rowCount }, () => `(${tuple})`).join(', ');\n  return { sql: `${head}${rows}${tail}`, width };\n}\n\nexport function rewriteUpdate(sql: string, rowCount: number): { sql: string; width: number } | null {\n  const m = UPDATE_RE.exec(sql.trim());\n  if (!m) return null;\n  const [, table, setClause, whereClause] = m;\n  const setCols = equalityColumns(setClause, /,/);\n  const whereCols = equalityColumns(whereClause, /\\bAND\\b/i);\n  if (!setCols || !whereCols) return null;\n  const dataCols = [...setCols, ...whereCols];\n  const setAssign = setCols.map((c) => `${c} = data.${c}`).join(', ');\n  const joinCond = whereCols.map((c) => `${table}.${c} = data.${c}`).join(' AND ');\n  return { sql: `UPDATE ${table} SET ${setAssign} FROM (VALUES ${tuples(rowCount, dataCols.length)}) AS data(${dataCols.join(', ')}) WHERE ${joinCond}`, width: dataCols.length };\n}\n\nexport function rewriteDelete(sql: string, rowCount: number): { sql: string; width: number } | null {\n  const m = DELETE_RE.exec(sql.trim());\n  if (!m) return null;\n  const [, table, whereClause] = m;\n  const cols = equalityColumns(whereClause, /\\bAND\\b/i);\n  if (!cols) return null;\n  if (cols.length === 1) {\n    return { sql: `DELETE FROM ${table} WHERE ${cols[0]} IN (${Array.from({ length: rowCount }, () => '?').join(', ')})`, width: 1 };\n  }\n  return { sql: `DELETE FROM ${table} WHERE (${cols.join(', ')}) IN (VALUES ${tuples(rowCount, cols.length)})`, width: cols.length };\n}\n\n// Tries each shape in turn; null means the caller should fall back to a per-row loop.\nexport function rewriteBatch(sql: string, rowCount: number): { sql: string; width: number } | null {\n  return rewriteInsert(sql, rowCount) ?? rewriteUpdate(sql, rowCount) ?? rewriteDelete(sql, rowCount);\n}\n"],"names":["rewriteBatch","rewriteDelete","rewriteInsert","rewriteUpdate","INSERT_RE","UPDATE_RE","DELETE_RE","placeholderCount","tuple","match","length","equalityColumns","clause","sep","parts","split","map","p","trim","cols","part","m","exec","push","tuples","rowCount","width","row","Array","from","join","sql","head","tail","rows","table","setClause","whereClause","setCols","whereCols","dataCols","setAssign","c","joinCond"],"mappings":";;;;;;;;;;;QAiEgBA;eAAAA;;QAbAC;eAAAA;;QAvBAC;eAAAA;;QAUAC;eAAAA;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;AAvChB,qGAAqG;AACrG,yHAAyH;AAEzH,IAAMC,YAAY;AAClB,IAAMC,YAAY;AAClB,IAAMC,YAAY;AAElB,SAASC,iBAAiBC,KAAa;QAC7BA;IAAR,OAAO,EAACA,eAAAA,MAAMC,KAAK,CAAC,oBAAZD,0BAAAA,eAAsB,EAAE,EAAEE,MAAM;AAC1C;AAEA,gGAAgG;AAChG,sEAAsE;AACtE,SAASC,gBAAgBC,MAAc,EAAEC,GAAW;IAClD,IAAMC,QAAQF,OAAOG,KAAK,CAACF,KAAKG,GAAG,CAAC,SAACC;eAAMA,EAAEC,IAAI;;IACjD,IAAMC,OAAiB,EAAE;QACpB,kCAAA,2BAAA;;QAAL,QAAK,YAAcL,0BAAd,SAAA,6BAAA,QAAA,yBAAA,iCAAqB;YAArB,IAAMM,OAAN;YACH,IAAMC,IAAI,gBAAgBC,IAAI,CAACF;YAC/B,IAAI,CAACC,GAAG,OAAO;YACfF,KAAKI,IAAI,CAACF,CAAC,CAAC,EAAE,CAACH,IAAI;QACrB;;QAJK;QAAA;;;iBAAA,6BAAA;gBAAA;;;gBAAA;sBAAA;;;;IAKL,OAAOC;AACT;AAEA,SAASK,OAAOC,QAAgB,EAAEC,KAAa;IAC7C,IAAMC,MAAM,AAAC,IAAuD,OAApDC,MAAMC,IAAI,CAAC;QAAEnB,QAAQgB;IAAM,GAAG;eAAM;OAAKI,IAAI,CAAC,OAAM;IACpE,OAAOF,MAAMC,IAAI,CAAC;QAAEnB,QAAQe;IAAS,GAAG;eAAME;OAAKG,IAAI,CAAC;AAC1D;AAEO,SAAS5B,cAAc6B,GAAW,EAAEN,QAAgB;IACzD,IAAMJ,IAAIjB,UAAUkB,IAAI,CAACS,IAAIb,IAAI;IACjC,IAAI,CAACG,GAAG,OAAO;IACf,IAA8BA,sBAAAA,OAArBW,OAAqBX,OAAfb,QAAea,OAARY,OAAQZ;IAC9B,IAAMK,QAAQnB,iBAAiBC;IAC/B,IAAIkB,UAAU,GAAG,OAAO;IACxB,IAAMQ,OAAON,MAAMC,IAAI,CAAC;QAAEnB,QAAQe;IAAS,GAAG;eAAM,AAAC,IAAS,OAANjB,OAAM;OAAIsB,IAAI,CAAC;IACvE,OAAO;QAAEC,KAAK,AAAC,GAASG,OAAPF,MAAcC,OAAPC,MAAY,OAALD;QAAQP,OAAAA;IAAM;AAC/C;AAEO,SAASvB,cAAc4B,GAAW,EAAEN,QAAgB;IACzD,IAAMJ,IAAIhB,UAAUiB,IAAI,CAACS,IAAIb,IAAI;IACjC,IAAI,CAACG,GAAG,OAAO;IACf,IAA0CA,sBAAAA,OAAjCc,QAAiCd,OAA1Be,YAA0Bf,OAAfgB,cAAehB;IAC1C,IAAMiB,UAAU3B,gBAAgByB,WAAW;IAC3C,IAAMG,YAAY5B,gBAAgB0B,aAAa;IAC/C,IAAI,CAACC,WAAW,CAACC,WAAW,OAAO;IACnC,IAAMC,WAAW,AAAC,qBAAGF,gBAAS,qBAAGC;IACjC,IAAME,YAAYH,QAAQtB,GAAG,CAAC,SAAC0B;eAAM,AAAC,GAAcA,OAAZA,GAAE,YAAY,OAAFA;OAAKZ,IAAI,CAAC;IAC9D,IAAMa,WAAWJ,UAAUvB,GAAG,CAAC,SAAC0B;eAAM,AAAC,GAAWA,OAATP,OAAM,KAAeO,OAAZA,GAAE,YAAY,OAAFA;OAAKZ,IAAI,CAAC;IACxE,OAAO;QAAEC,KAAK,AAAC,UAAsBU,OAAbN,OAAM,SAAiCX,OAA1BiB,WAAU,kBAA8DD,OAA9ChB,OAAOC,UAAUe,SAAS9B,MAAM,GAAE,cAA0CiC,OAA9BH,SAASV,IAAI,CAAC,OAAM,YAAmB,OAATa;QAAYjB,OAAOc,SAAS9B,MAAM;IAAC;AAChL;AAEO,SAAST,cAAc8B,GAAW,EAAEN,QAAgB;IACzD,IAAMJ,IAAIf,UAAUgB,IAAI,CAACS,IAAIb,IAAI;IACjC,IAAI,CAACG,GAAG,OAAO;IACf,IAA+BA,sBAAAA,OAAtBc,QAAsBd,OAAfgB,cAAehB;IAC/B,IAAMF,OAAOR,gBAAgB0B,aAAa;IAC1C,IAAI,CAAClB,MAAM,OAAO;IAClB,IAAIA,KAAKT,MAAM,KAAK,GAAG;QACrB,OAAO;YAAEqB,KAAK,AAAC,eAA6BZ,OAAfgB,OAAM,WAAwBP,OAAfT,IAAI,CAAC,EAAE,EAAC,SAA8D,OAAvDS,MAAMC,IAAI,CAAC;gBAAEnB,QAAQe;YAAS,GAAG;uBAAM;eAAKK,IAAI,CAAC,OAAM;YAAIJ,OAAO;QAAE;IACjI;IACA,OAAO;QAAEK,KAAK,AAAC,eAA8BZ,OAAhBgB,OAAM,YAAyCX,OAA/BL,KAAKW,IAAI,CAAC,OAAM,iBAA6C,OAA9BN,OAAOC,UAAUN,KAAKT,MAAM,GAAE;QAAIgB,OAAOP,KAAKT,MAAM;IAAC;AACnI;AAGO,SAASV,aAAa+B,GAAW,EAAEN,QAAgB;QACjDvB,MAAAA;IAAP,QAAOA,QAAAA,iBAAAA,cAAc6B,KAAKN,uBAAnBvB,4BAAAA,iBAAgCC,cAAc4B,KAAKN,uBAAnDvB,kBAAAA,OAAgED,cAAc8B,KAAKN;AAC5F"}
+{"version":3,"sources":["/Users/kevin/Dev/OpenSource/ai/sensemaking/src/store/duckdb/batch.ts"],"sourcesContent":["// Rewrites the store's single-row parameterized INSERT/UPDATE/DELETE into one multi-row statement --\n// DuckDB's columnar engine pays for row-at-a-time writes, not just the round trip (PRINCIPLES: documented-means-tested).\n// rewriteInsert itself lives in the shared ../batch.ts: turso folds INSERTs the same way (3.36 B).\n\nimport { rewriteInsert } from '../batch.ts';\n\nconst UPDATE_RE = /^UPDATE\\s+(\\S+)\\s+SET\\s+(.+?)\\s+WHERE\\s+(.+)$/is;\nconst DELETE_RE = /^DELETE\\s+FROM\\s+(\\S+)\\s+WHERE\\s+(.+)$/is;\n\n// \"a = ?, b = ?\" or \"a = ? AND b = ?\" -> [\"a\", \"b\"]. A clause that doesn't fit yields a shorter\n// list, which rewriteUpdate/rewriteDelete reject as a width mismatch.\nfunction equalityColumns(clause: string, sep: RegExp): string[] | null {\n  const parts = clause.split(sep).map((p) => p.trim());\n  const cols: string[] = [];\n  for (const part of parts) {\n    const m = /^(.+?)=\\s*\\?$/.exec(part);\n    if (!m) return null;\n    cols.push(m[1].trim());\n  }\n  return cols;\n}\n\nfunction tuples(rowCount: number, width: number): string {\n  const row = `(${Array.from({ length: width }, () => '?').join(', ')})`;\n  return Array.from({ length: rowCount }, () => row).join(', ');\n}\n\nexport function rewriteUpdate(sql: string, rowCount: number): { sql: string; width: number } | null {\n  const m = UPDATE_RE.exec(sql.trim());\n  if (!m) return null;\n  const [, table, setClause, whereClause] = m;\n  const setCols = equalityColumns(setClause, /,/);\n  const whereCols = equalityColumns(whereClause, /\\bAND\\b/i);\n  if (!setCols || !whereCols) return null;\n  const dataCols = [...setCols, ...whereCols];\n  const setAssign = setCols.map((c) => `${c} = data.${c}`).join(', ');\n  const joinCond = whereCols.map((c) => `${table}.${c} = data.${c}`).join(' AND ');\n  return { sql: `UPDATE ${table} SET ${setAssign} FROM (VALUES ${tuples(rowCount, dataCols.length)}) AS data(${dataCols.join(', ')}) WHERE ${joinCond}`, width: dataCols.length };\n}\n\nexport function rewriteDelete(sql: string, rowCount: number): { sql: string; width: number } | null {\n  const m = DELETE_RE.exec(sql.trim());\n  if (!m) return null;\n  const [, table, whereClause] = m;\n  const cols = equalityColumns(whereClause, /\\bAND\\b/i);\n  if (!cols) return null;\n  if (cols.length === 1) {\n    return { sql: `DELETE FROM ${table} WHERE ${cols[0]} IN (${Array.from({ length: rowCount }, () => '?').join(', ')})`, width: 1 };\n  }\n  return { sql: `DELETE FROM ${table} WHERE (${cols.join(', ')}) IN (VALUES ${tuples(rowCount, cols.length)})`, width: cols.length };\n}\n\n// Tries each shape in turn; null means the caller should fall back to a per-row loop.\nexport function rewriteBatch(sql: string, rowCount: number): { sql: string; width: number } | null {\n  return rewriteInsert(sql, rowCount) ?? rewriteUpdate(sql, rowCount) ?? rewriteDelete(sql, rowCount);\n}\n"],"names":["rewriteBatch","rewriteDelete","rewriteUpdate","UPDATE_RE","DELETE_RE","equalityColumns","clause","sep","parts","split","map","p","trim","cols","part","m","exec","push","tuples","rowCount","width","row","Array","from","length","join","sql","table","setClause","whereClause","setCols","whereCols","dataCols","setAssign","c","joinCond","rewriteInsert"],"mappings":"AAAA,qGAAqG;AACrG,yHAAyH;AACzH,mGAAmG;;;;;;;;;;;;QAmDnFA;eAAAA;;QAbAC;eAAAA;;QAbAC;eAAAA;;;uBAvBc;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;AAE9B,IAAMC,YAAY;AAClB,IAAMC,YAAY;AAElB,gGAAgG;AAChG,sEAAsE;AACtE,SAASC,gBAAgBC,MAAc,EAAEC,GAAW;IAClD,IAAMC,QAAQF,OAAOG,KAAK,CAACF,KAAKG,GAAG,CAAC,SAACC;eAAMA,EAAEC,IAAI;;IACjD,IAAMC,OAAiB,EAAE;QACpB,kCAAA,2BAAA;;QAAL,QAAK,YAAcL,0BAAd,SAAA,6BAAA,QAAA,yBAAA,iCAAqB;YAArB,IAAMM,OAAN;YACH,IAAMC,IAAI,gBAAgBC,IAAI,CAACF;YAC/B,IAAI,CAACC,GAAG,OAAO;YACfF,KAAKI,IAAI,CAACF,CAAC,CAAC,EAAE,CAACH,IAAI;QACrB;;QAJK;QAAA;;;iBAAA,6BAAA;gBAAA;;;gBAAA;sBAAA;;;;IAKL,OAAOC;AACT;AAEA,SAASK,OAAOC,QAAgB,EAAEC,KAAa;IAC7C,IAAMC,MAAM,AAAC,IAAuD,OAApDC,MAAMC,IAAI,CAAC;QAAEC,QAAQJ;IAAM,GAAG;eAAM;OAAKK,IAAI,CAAC,OAAM;IACpE,OAAOH,MAAMC,IAAI,CAAC;QAAEC,QAAQL;IAAS,GAAG;eAAME;OAAKI,IAAI,CAAC;AAC1D;AAEO,SAASvB,cAAcwB,GAAW,EAAEP,QAAgB;IACzD,IAAMJ,IAAIZ,UAAUa,IAAI,CAACU,IAAId,IAAI;IACjC,IAAI,CAACG,GAAG,OAAO;IACf,IAA0CA,sBAAAA,OAAjCY,QAAiCZ,OAA1Ba,YAA0Bb,OAAfc,cAAed;IAC1C,IAAMe,UAAUzB,gBAAgBuB,WAAW;IAC3C,IAAMG,YAAY1B,gBAAgBwB,aAAa;IAC/C,IAAI,CAACC,WAAW,CAACC,WAAW,OAAO;IACnC,IAAMC,WAAW,AAAC,qBAAGF,gBAAS,qBAAGC;IACjC,IAAME,YAAYH,QAAQpB,GAAG,CAAC,SAACwB;eAAM,AAAC,GAAcA,OAAZA,GAAE,YAAY,OAAFA;OAAKT,IAAI,CAAC;IAC9D,IAAMU,WAAWJ,UAAUrB,GAAG,CAAC,SAACwB;eAAM,AAAC,GAAWA,OAATP,OAAM,KAAeO,OAAZA,GAAE,YAAY,OAAFA;OAAKT,IAAI,CAAC;IACxE,OAAO;QAAEC,KAAK,AAAC,UAAsBO,OAAbN,OAAM,SAAiCT,OAA1Be,WAAU,kBAA8DD,OAA9Cd,OAAOC,UAAUa,SAASR,MAAM,GAAE,cAA0CW,OAA9BH,SAASP,IAAI,CAAC,OAAM,YAAmB,OAATU;QAAYf,OAAOY,SAASR,MAAM;IAAC;AAChL;AAEO,SAASvB,cAAcyB,GAAW,EAAEP,QAAgB;IACzD,IAAMJ,IAAIX,UAAUY,IAAI,CAACU,IAAId,IAAI;IACjC,IAAI,CAACG,GAAG,OAAO;IACf,IAA+BA,sBAAAA,OAAtBY,QAAsBZ,OAAfc,cAAed;IAC/B,IAAMF,OAAOR,gBAAgBwB,aAAa;IAC1C,IAAI,CAAChB,MAAM,OAAO;IAClB,IAAIA,KAAKW,MAAM,KAAK,GAAG;QACrB,OAAO;YAAEE,KAAK,AAAC,eAA6Bb,OAAfc,OAAM,WAAwBL,OAAfT,IAAI,CAAC,EAAE,EAAC,SAA8D,OAAvDS,MAAMC,IAAI,CAAC;gBAAEC,QAAQL;YAAS,GAAG;uBAAM;eAAKM,IAAI,CAAC,OAAM;YAAIL,OAAO;QAAE;IACjI;IACA,OAAO;QAAEM,KAAK,AAAC,eAA8Bb,OAAhBc,OAAM,YAAyCT,OAA/BL,KAAKY,IAAI,CAAC,OAAM,iBAA6C,OAA9BP,OAAOC,UAAUN,KAAKW,MAAM,GAAE;QAAIJ,OAAOP,KAAKW,MAAM;IAAC;AACnI;AAGO,SAASxB,aAAa0B,GAAW,EAAEP,QAAgB;QACjDiB,MAAAA;IAAP,QAAOA,QAAAA,iBAAAA,IAAAA,sBAAa,EAACV,KAAKP,uBAAnBiB,4BAAAA,iBAAgClC,cAAcwB,KAAKP,uBAAnDiB,kBAAAA,OAAgEnC,cAAcyB,KAAKP;AAC5F"}

package/dist/cjs/store/duckdb/fieldStats.js

@@ -202,13 +202,10 @@ function classifyVariantType(name, field) {
     if (name === 'VARCHAR') return 'text';
     throw new Error('duckdb: variant_typeof() returned unrecognized type "'.concat(name, '" for column "').concat(field, '"'));
 }
-// duckdb degrades superlinearly in the number of string_agg(DISTINCT variant_typeof()) aggregates in
-// one projection: 301 columns cost 1148ms in one query against 10ms for COUNT alone over the same
-// columns, so it is the aggregate count, not the width. Measured chunk sizes at 301 columns:
-// 10 -> 71ms, 16 -> 81ms, 25 -> 88ms, 40 -> 110ms, 64 -> 158ms, flat between 10 and 25. Chunking
-// also wins on an ordinary tree (31 columns: 17ms in one query, 11ms chunked), so there is no
-// small-tree cost to trade against. sqlite and turso need none of this: GROUP_CONCAT over 300
-// columns runs in 7ms.
+// duckdb degrades superlinearly in the number of string_agg(DISTINCT variant_typeof()) aggregates
+// per projection, not the column width (301 columns: 1148ms unchunked vs 88ms at chunk size 25);
+// 16 sits inside the measured flat 10-25 range. sqlite/turso need none of this: GROUP_CONCAT over
+// 300 columns runs in 7ms.
 var AGGREGATES_PER_QUERY = 16;
 function fieldStats(conn, columns, scopeWhere) {
     return _async_to_generator(function() {

package/dist/cjs/store/duckdb/fieldStats.js.map

@@ -1 +1 @@
-{"version":3,"sources":["/Users/kevin/Dev/OpenSource/ai/sensemaking/src/store/duckdb/fieldStats.ts"],"sourcesContent":["import { quoteIdent } from '../shared.ts';\nimport type { Connection, FieldStat } from '../types.ts';\n\n// DuckDB's frontmatter columns are VARIANT (reconcile.ts), so typeof() always reports VARIANT; variant_typeof() reports the boxed native\n// type instead -- INT128, DOUBLE, VARCHAR, or a literal's narrower type like DECIMAL(p, s) -- mapped onto sqlite/turso's integer/real/text vocabulary; an unrecognized name throws (no-silent-modes).\nfunction classifyVariantType(name: string, field: string): 'integer' | 'real' | 'text' {\n  if (/^U?INT\\d+$/.test(name)) return 'integer';\n  if (name === 'FLOAT' || name === 'DOUBLE' || name.startsWith('DECIMAL(')) return 'real';\n  if (name === 'VARCHAR') return 'text';\n  throw new Error(`duckdb: variant_typeof() returned unrecognized type \"${name}\" for column \"${field}\"`);\n}\n\n// duckdb degrades superlinearly in the number of string_agg(DISTINCT variant_typeof()) aggregates in\n// one projection: 301 columns cost 1148ms in one query against 10ms for COUNT alone over the same\n// columns, so it is the aggregate count, not the width. Measured chunk sizes at 301 columns:\n// 10 -> 71ms, 16 -> 81ms, 25 -> 88ms, 40 -> 110ms, 64 -> 158ms, flat between 10 and 25. Chunking\n// also wins on an ordinary tree (31 columns: 17ms in one query, 11ms chunked), so there is no\n// small-tree cost to trade against. sqlite and turso need none of this: GROUP_CONCAT over 300\n// columns runs in 7ms.\nconst AGGREGATES_PER_QUERY = 16;\n\n// COUNT + string_agg(DISTINCT variant_typeof()) per column keeps this O(columns), not O(rows x columns).\n// '|' is the separator, not ',': DECIMAL's display name (\"DECIMAL(4, 2)\") contains a comma.\nexport async function fieldStats(conn: Connection, columns: string[], scopeWhere: string): Promise<FieldStat[]> {\n  const stats: FieldStat[] = [];\n  for (let start = 0; start < columns.length; start += AGGREGATES_PER_QUERY) {\n    const chunk = columns.slice(start, start + AGGREGATES_PER_QUERY);\n    const exprs = chunk.map((name, i) => {\n      const quoted = quoteIdent(name);\n      return `COUNT(${quoted}) AS n${i}, string_agg(DISTINCT variant_typeof(${quoted}), '|') FILTER (WHERE ${quoted} IS NOT NULL) AS t${i}`;\n    });\n    const stmt = await conn.prepare(`SELECT ${exprs.join(', ')} FROM frontmatter ${scopeWhere}`);\n    const row = (await stmt.get()) as Record<string, number | string | null>;\n    for (const [i, name] of chunk.entries()) {\n      const raw = ((row[`t${i}`] as string) ?? '').split('|').filter(Boolean);\n      const types = [...new Set(raw.map((t) => classifyVariantType(t, name)))].sort();\n      stats.push({ field: name, coverage: Number(row[`n${i}`] ?? 0), type: types.join(',') });\n    }\n  }\n  return stats;\n}\n"],"names":["fieldStats","classifyVariantType","name","field","test","startsWith","Error","AGGREGATES_PER_QUERY","conn","columns","scopeWhere","stats","start","chunk","exprs","stmt","row","length","slice","map","i","quoted","quoteIdent","prepare","join","get","raw","split","filter","Boolean","types","Set","t","sort","push","coverage","Number","type","entries"],"mappings":";;;;+BAuBsBA;;;eAAAA;;;wBAvBK;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;AAG3B,yIAAyI;AACzI,sMAAsM;AACtM,SAASC,oBAAoBC,IAAY,EAAEC,KAAa;IACtD,IAAI,aAAaC,IAAI,CAACF,OAAO,OAAO;IACpC,IAAIA,SAAS,WAAWA,SAAS,YAAYA,KAAKG,UAAU,CAAC,aAAa,OAAO;IACjF,IAAIH,SAAS,WAAW,OAAO;IAC/B,MAAM,IAAII,MAAM,AAAC,wDAA4EH,OAArBD,MAAK,kBAAsB,OAANC,OAAM;AACrG;AAEA,qGAAqG;AACrG,kGAAkG;AAClG,6FAA6F;AAC7F,iGAAiG;AACjG,8FAA8F;AAC9F,8FAA8F;AAC9F,uBAAuB;AACvB,IAAMI,uBAAuB;AAItB,SAAeP,WAAWQ,IAAgB,EAAEC,OAAiB,EAAEC,UAAkB;;YAChFC,OACGC,OACDC,OACAC,OAIAC,MACAC,KACD,2BAAA,mBAAA,uBAAA,WAAA;;;;oBATDL;oBACGC,QAAQ;;;yBAAGA,CAAAA,QAAQH,QAAQQ,MAAM,AAAD;;;;oBACjCJ,QAAQJ,QAAQS,KAAK,CAACN,OAAOA,QAAQL;oBACrCO,QAAQD,MAAMM,GAAG,CAAC,SAACjB,MAAMkB;wBAC7B,IAAMC,SAASC,IAAAA,oBAAU,EAACpB;wBAC1B,OAAO,AAAC,SAAuBkB,OAAfC,QAAO,UAAiDA,OAAzCD,GAAE,yCAAsEC,OAA/BA,QAAO,0BAAmDD,OAA3BC,QAAO,sBAAsB,OAAFD;oBACpI;oBACa;;wBAAMZ,KAAKe,OAAO,CAAC,AAAC,UAA8Cb,OAArCI,MAAMU,IAAI,CAAC,OAAM,sBAA+B,OAAXd;;;oBAAzEK,OAAO;oBACA;;wBAAMA,KAAKU,GAAG;;;oBAArBT,MAAO;oBACR,kCAAA,2BAAA;;;4BAAA,mCAAA,iBAAOI,oBAAGlB;gCACCc,OAE6BA;4BAF3C,IAAMU,MAAM,EAAEV,QAAAA,GAAG,CAAC,AAAC,IAAK,OAAFI,GAAI,cAAZJ,mBAAAA,QAA2B,IAAIW,KAAK,CAAC,KAAKC,MAAM,CAACC;4BAC/D,IAAMC,QAAQ,AAAC,qBAAG,IAAIC,IAAIL,IAAIP,GAAG,CAAC,SAACa;uCAAM/B,oBAAoB+B,GAAG9B;iCAAS+B,IAAI;4BAC7EtB,MAAMuB,IAAI,CAAC;gCAAE/B,OAAOD;gCAAMiC,UAAUC,QAAOpB,SAAAA,GAAG,CAAC,AAAC,IAAK,OAAFI,GAAI,cAAZJ,oBAAAA,SAAgB;gCAAIqB,MAAMP,MAAMN,IAAI,CAAC;4BAAK;wBACvF;wBAJA,IAAK,YAAmBX,MAAMyB,OAAO,yBAAhC,6BAAA,QAAA,yBAAA;;wBAAA;wBAAA;;;iCAAA,6BAAA;gCAAA;;;gCAAA;sCAAA;;;;;;oBARqC1B,SAASL;;;;;;oBAcrD;;wBAAOI;;;;IACT"}
+{"version":3,"sources":["/Users/kevin/Dev/OpenSource/ai/sensemaking/src/store/duckdb/fieldStats.ts"],"sourcesContent":["import { quoteIdent } from '../shared.ts';\nimport type { Connection, FieldStat } from '../types.ts';\n\n// DuckDB's frontmatter columns are VARIANT (reconcile.ts), so typeof() always reports VARIANT; variant_typeof() reports the boxed native\n// type instead -- INT128, DOUBLE, VARCHAR, or a literal's narrower type like DECIMAL(p, s) -- mapped onto sqlite/turso's integer/real/text vocabulary; an unrecognized name throws (no-silent-modes).\nfunction classifyVariantType(name: string, field: string): 'integer' | 'real' | 'text' {\n  if (/^U?INT\\d+$/.test(name)) return 'integer';\n  if (name === 'FLOAT' || name === 'DOUBLE' || name.startsWith('DECIMAL(')) return 'real';\n  if (name === 'VARCHAR') return 'text';\n  throw new Error(`duckdb: variant_typeof() returned unrecognized type \"${name}\" for column \"${field}\"`);\n}\n\n// duckdb degrades superlinearly in the number of string_agg(DISTINCT variant_typeof()) aggregates\n// per projection, not the column width (301 columns: 1148ms unchunked vs 88ms at chunk size 25);\n// 16 sits inside the measured flat 10-25 range. sqlite/turso need none of this: GROUP_CONCAT over\n// 300 columns runs in 7ms.\nconst AGGREGATES_PER_QUERY = 16;\n\n// COUNT + string_agg(DISTINCT variant_typeof()) per column keeps this O(columns), not O(rows x columns).\n// '|' is the separator, not ',': DECIMAL's display name (\"DECIMAL(4, 2)\") contains a comma.\nexport async function fieldStats(conn: Connection, columns: string[], scopeWhere: string): Promise<FieldStat[]> {\n  const stats: FieldStat[] = [];\n  for (let start = 0; start < columns.length; start += AGGREGATES_PER_QUERY) {\n    const chunk = columns.slice(start, start + AGGREGATES_PER_QUERY);\n    const exprs = chunk.map((name, i) => {\n      const quoted = quoteIdent(name);\n      return `COUNT(${quoted}) AS n${i}, string_agg(DISTINCT variant_typeof(${quoted}), '|') FILTER (WHERE ${quoted} IS NOT NULL) AS t${i}`;\n    });\n    const stmt = await conn.prepare(`SELECT ${exprs.join(', ')} FROM frontmatter ${scopeWhere}`);\n    const row = (await stmt.get()) as Record<string, number | string | null>;\n    for (const [i, name] of chunk.entries()) {\n      const raw = ((row[`t${i}`] as string) ?? '').split('|').filter(Boolean);\n      const types = [...new Set(raw.map((t) => classifyVariantType(t, name)))].sort();\n      stats.push({ field: name, coverage: Number(row[`n${i}`] ?? 0), type: types.join(',') });\n    }\n  }\n  return stats;\n}\n"],"names":["fieldStats","classifyVariantType","name","field","test","startsWith","Error","AGGREGATES_PER_QUERY","conn","columns","scopeWhere","stats","start","chunk","exprs","stmt","row","length","slice","map","i","quoted","quoteIdent","prepare","join","get","raw","split","filter","Boolean","types","Set","t","sort","push","coverage","Number","type","entries"],"mappings":";;;;+BAoBsBA;;;eAAAA;;;wBApBK;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;AAG3B,yIAAyI;AACzI,sMAAsM;AACtM,SAASC,oBAAoBC,IAAY,EAAEC,KAAa;IACtD,IAAI,aAAaC,IAAI,CAACF,OAAO,OAAO;IACpC,IAAIA,SAAS,WAAWA,SAAS,YAAYA,KAAKG,UAAU,CAAC,aAAa,OAAO;IACjF,IAAIH,SAAS,WAAW,OAAO;IAC/B,MAAM,IAAII,MAAM,AAAC,wDAA4EH,OAArBD,MAAK,kBAAsB,OAANC,OAAM;AACrG;AAEA,kGAAkG;AAClG,iGAAiG;AACjG,kGAAkG;AAClG,2BAA2B;AAC3B,IAAMI,uBAAuB;AAItB,SAAeP,WAAWQ,IAAgB,EAAEC,OAAiB,EAAEC,UAAkB;;YAChFC,OACGC,OACDC,OACAC,OAIAC,MACAC,KACD,2BAAA,mBAAA,uBAAA,WAAA;;;;oBATDL;oBACGC,QAAQ;;;yBAAGA,CAAAA,QAAQH,QAAQQ,MAAM,AAAD;;;;oBACjCJ,QAAQJ,QAAQS,KAAK,CAACN,OAAOA,QAAQL;oBACrCO,QAAQD,MAAMM,GAAG,CAAC,SAACjB,MAAMkB;wBAC7B,IAAMC,SAASC,IAAAA,oBAAU,EAACpB;wBAC1B,OAAO,AAAC,SAAuBkB,OAAfC,QAAO,UAAiDA,OAAzCD,GAAE,yCAAsEC,OAA/BA,QAAO,0BAAmDD,OAA3BC,QAAO,sBAAsB,OAAFD;oBACpI;oBACa;;wBAAMZ,KAAKe,OAAO,CAAC,AAAC,UAA8Cb,OAArCI,MAAMU,IAAI,CAAC,OAAM,sBAA+B,OAAXd;;;oBAAzEK,OAAO;oBACA;;wBAAMA,KAAKU,GAAG;;;oBAArBT,MAAO;oBACR,kCAAA,2BAAA;;;4BAAA,mCAAA,iBAAOI,oBAAGlB;gCACCc,OAE6BA;4BAF3C,IAAMU,MAAM,EAAEV,QAAAA,GAAG,CAAC,AAAC,IAAK,OAAFI,GAAI,cAAZJ,mBAAAA,QAA2B,IAAIW,KAAK,CAAC,KAAKC,MAAM,CAACC;4BAC/D,IAAMC,QAAQ,AAAC,qBAAG,IAAIC,IAAIL,IAAIP,GAAG,CAAC,SAACa;uCAAM/B,oBAAoB+B,GAAG9B;iCAAS+B,IAAI;4BAC7EtB,MAAMuB,IAAI,CAAC;gCAAE/B,OAAOD;gCAAMiC,UAAUC,QAAOpB,SAAAA,GAAG,CAAC,AAAC,IAAK,OAAFI,GAAI,cAAZJ,oBAAAA,SAAgB;gCAAIqB,MAAMP,MAAMN,IAAI,CAAC;4BAAK;wBACvF;wBAJA,IAAK,YAAmBX,MAAMyB,OAAO,yBAAhC,6BAAA,QAAA,yBAAA;;wBAAA;wBAAA;;;iCAAA,6BAAA;gCAAA;;;gCAAA;sCAAA;;;;;;oBARqC1B,SAASL;;;;;;oBAcrD;;wBAAOI;;;;IACT"}

package/dist/cjs/store/duckdb/lexical.d.cts

@@ -2,10 +2,10 @@ import type { Connection, LexicalHit, LexicalQueryOptions } from '../types.js';
 interface FtsIndexState {
     stale: boolean;
 }
-export declare function markContentStale(conn: Connection): void;
+export declare function markContentStale(conn: Connection): Promise<void>;
 export declare function queryLexical(conn: Connection, terms: string, opts: LexicalQueryOptions, state: FtsIndexState): Promise<LexicalHit[]>;
 export declare function createLexicalIndex(conn: Connection): {
     query: (terms: string, opts: LexicalQueryOptions) => Promise<LexicalHit[]>;
-    markStale: () => void;
+    markStale: () => Promise<void>;
 };
 export {};

package/dist/cjs/store/duckdb/lexical.d.ts

@@ -2,10 +2,10 @@ import type { Connection, LexicalHit, LexicalQueryOptions } from '../types.js';
 interface FtsIndexState {
     stale: boolean;
 }
-export declare function markContentStale(conn: Connection): void;
+export declare function markContentStale(conn: Connection): Promise<void>;
 export declare function queryLexical(conn: Connection, terms: string, opts: LexicalQueryOptions, state: FtsIndexState): Promise<LexicalHit[]>;
 export declare function createLexicalIndex(conn: Connection): {
     query: (terms: string, opts: LexicalQueryOptions) => Promise<LexicalHit[]>;
-    markStale: () => void;
+    markStale: () => Promise<void>;
 };
 export {};