wiki-graph-core 0.3.1 → 0.4.0

package/dist/source/meta.js.map

@@ -1 +0,0 @@
-{"version":3,"sources":["../../src/source/meta.ts"],"sourcesContent":["import { z } from \"zod\";\n\nexport const SOURCE_FORMATS = [\"epub\", \"pdf\", \"txt\", \"markdown\"] as const;\nexport const BOOK_META_VERSION = 1;\n\nexport const sourceFormatSchema = z.enum(SOURCE_FORMATS);\n\nexport const bookMetaSchema = z.object({\n  version: z.literal(BOOK_META_VERSION),\n  sourceFormat: sourceFormatSchema,\n  title: z.string().min(1).nullable(),\n  authors: z.array(z.string().min(1)),\n  language: z.string().min(1).nullable(),\n  identifier: z.string().min(1).nullable(),\n  publisher: z.string().min(1).nullable(),\n  publishedAt: z.string().min(1).nullable(),\n  description: z.string().min(1).nullable(),\n});\n\nexport type SourceFormat = z.infer<typeof sourceFormatSchema>;\nexport type BookMeta = z.infer<typeof bookMetaSchema>;\n"],"mappings":"AAAA,SAAS,SAAS;AAEX,MAAM,iBAAiB,CAAC,QAAQ,OAAO,OAAO,UAAU;AACxD,MAAM,oBAAoB;AAE1B,MAAM,qBAAqB,EAAE,KAAK,cAAc;AAEhD,MAAM,iBAAiB,EAAE,OAAO;AAAA,EACrC,SAAS,EAAE,QAAQ,iBAAiB;AAAA,EACpC,cAAc;AAAA,EACd,OAAO,EAAE,OAAO,EAAE,IAAI,CAAC,EAAE,SAAS;AAAA,EAClC,SAAS,EAAE,MAAM,EAAE,OAAO,EAAE,IAAI,CAAC,CAAC;AAAA,EAClC,UAAU,EAAE,OAAO,EAAE,IAAI,CAAC,EAAE,SAAS;AAAA,EACrC,YAAY,EAAE,OAAO,EAAE,IAAI,CAAC,EAAE,SAAS;AAAA,EACvC,WAAW,EAAE,OAAO,EAAE,IAAI,CAAC,EAAE,SAAS;AAAA,EACtC,aAAa,EAAE,OAAO,EAAE,IAAI,CAAC,EAAE,SAAS;AAAA,EACxC,aAAa,EAAE,OAAO,EAAE,IAAI,CAAC,EAAE,SAAS;AAC1C,CAAC;","names":[]}

package/dist/source/plain-text.js.map

@@ -1 +0,0 @@
-{"version":3,"sources":["../../src/source/plain-text.ts"],"sourcesContent":["import { createReadStream } from \"fs\";\nimport { stat } from \"fs/promises\";\nimport { basename, extname, resolve } from \"path\";\n\nimport { BOOK_META_VERSION, type BookMeta } from \"./meta.js\";\nimport type { SourceAdapter, SourceDocument } from \"./adapter.js\";\nimport type { SourceAsset, SourceSection, SourceTextStream } from \"./types.js\";\n\ntype PlainTextSourceFormat = \"markdown\" | \"txt\";\n\nconst ROOT_SECTION_ID = \"root\";\n\nclass PlainTextSection implements SourceSection {\n  readonly #path: string;\n  readonly #id: string;\n\n  public constructor(path: string, id = ROOT_SECTION_ID) {\n    this.#path = resolve(path);\n    this.#id = id;\n  }\n\n  public get id(): string {\n    return this.#id;\n  }\n\n  public get hasContent(): boolean {\n    return true;\n  }\n\n  public get title(): string | undefined {\n    return undefined;\n  }\n\n  public get children(): readonly SourceSection[] {\n    return [];\n  }\n\n  public open(): Promise<SourceTextStream> {\n    return Promise.resolve(createReadStream(this.#path, { encoding: \"utf8\" }));\n  }\n}\n\nclass PlainTextDocument implements SourceDocument {\n  readonly #path: string;\n  readonly #format: PlainTextSourceFormat;\n  readonly #section: PlainTextSection;\n\n  public constructor(path: string, format: PlainTextSourceFormat) {\n    this.#path = resolve(path);\n    this.#format = format;\n    this.#section = new PlainTextSection(this.#path);\n  }\n\n  public readMeta(): Promise<BookMeta> {\n    return Promise.resolve({\n      version: BOOK_META_VERSION,\n      sourceFormat: this.#format,\n      title: getFileStem(this.#path),\n      authors: [],\n      language: null,\n      identifier: null,\n      publisher: null,\n      publishedAt: null,\n      description: null,\n    });\n  }\n\n  public readCover(): Promise<SourceAsset | undefined> {\n    return Promise.resolve(undefined);\n  }\n\n  public readSections(): Promise<readonly SourceSection[]> {\n    return Promise.resolve([this.#section]);\n  }\n}\n\nexport class PlainTextSourceAdapter implements SourceAdapter {\n  readonly #format: PlainTextSourceFormat;\n\n  public constructor(format: PlainTextSourceFormat) {\n    this.#format = format;\n  }\n\n  public get format(): PlainTextSourceFormat {\n    return this.#format;\n  }\n\n  public async openSession<T>(\n    path: string,\n    operation: (document: SourceDocument) => Promise<T>,\n  ): Promise<T> {\n    const resolvedPath = resolve(path);\n    const fileStat = await stat(resolvedPath);\n\n    if (!fileStat.isFile()) {\n      throw new Error(`Source file is not a regular file: ${resolvedPath}`);\n    }\n\n    return await operation(new PlainTextDocument(resolvedPath, this.#format));\n  }\n}\n\nexport const TXT_SOURCE_ADAPTER = new PlainTextSourceAdapter(\"txt\");\nexport const MARKDOWN_SOURCE_ADAPTER = new PlainTextSourceAdapter(\"markdown\");\n\nfunction getFileStem(path: string): string | null {\n  const stem = basename(path, extname(path)).trim();\n\n  return stem === \"\" ? null : stem;\n}\n"],"mappings":"AAAA,SAAS,wBAAwB;AACjC,SAAS,YAAY;AACrB,SAAS,UAAU,SAAS,eAAe;AAE3C,SAAS,yBAAwC;AAMjD,MAAM,kBAAkB;AAExB,MAAM,iBAA0C;AAAA,EACrC;AAAA,EACA;AAAA,EAEF,YAAY,MAAc,KAAK,iBAAiB;AACrD,SAAK,QAAQ,QAAQ,IAAI;AACzB,SAAK,MAAM;AAAA,EACb;AAAA,EAEA,IAAW,KAAa;AACtB,WAAO,KAAK;AAAA,EACd;AAAA,EAEA,IAAW,aAAsB;AAC/B,WAAO;AAAA,EACT;AAAA,EAEA,IAAW,QAA4B;AACrC,WAAO;AAAA,EACT;AAAA,EAEA,IAAW,WAAqC;AAC9C,WAAO,CAAC;AAAA,EACV;AAAA,EAEO,OAAkC;AACvC,WAAO,QAAQ,QAAQ,iBAAiB,KAAK,OAAO,EAAE,UAAU,OAAO,CAAC,CAAC;AAAA,EAC3E;AACF;AAEA,MAAM,kBAA4C;AAAA,EACvC;AAAA,EACA;AAAA,EACA;AAAA,EAEF,YAAY,MAAc,QAA+B;AAC9D,SAAK,QAAQ,QAAQ,IAAI;AACzB,SAAK,UAAU;AACf,SAAK,WAAW,IAAI,iBAAiB,KAAK,KAAK;AAAA,EACjD;AAAA,EAEO,WAA8B;AACnC,WAAO,QAAQ,QAAQ;AAAA,MACrB,SAAS;AAAA,MACT,cAAc,KAAK;AAAA,MACnB,OAAO,YAAY,KAAK,KAAK;AAAA,MAC7B,SAAS,CAAC;AAAA,MACV,UAAU;AAAA,MACV,YAAY;AAAA,MACZ,WAAW;AAAA,MACX,aAAa;AAAA,MACb,aAAa;AAAA,IACf,CAAC;AAAA,EACH;AAAA,EAEO,YAA8C;AACnD,WAAO,QAAQ,QAAQ,MAAS;AAAA,EAClC;AAAA,EAEO,eAAkD;AACvD,WAAO,QAAQ,QAAQ,CAAC,KAAK,QAAQ,CAAC;AAAA,EACxC;AACF;AAEO,MAAM,uBAAgD;AAAA,EAClD;AAAA,EAEF,YAAY,QAA+B;AAChD,SAAK,UAAU;AAAA,EACjB;AAAA,EAEA,IAAW,SAAgC;AACzC,WAAO,KAAK;AAAA,EACd;AAAA,EAEA,MAAa,YACX,MACA,WACY;AACZ,UAAM,eAAe,QAAQ,IAAI;AACjC,UAAM,WAAW,MAAM,KAAK,YAAY;AAExC,QAAI,CAAC,SAAS,OAAO,GAAG;AACtB,YAAM,IAAI,MAAM,sCAAsC,YAAY,EAAE;AAAA,IACtE;AAEA,WAAO,MAAM,UAAU,IAAI,kBAAkB,cAAc,KAAK,OAAO,CAAC;AAAA,EAC1E;AACF;AAEO,MAAM,qBAAqB,IAAI,uBAAuB,KAAK;AAC3D,MAAM,0BAA0B,IAAI,uBAAuB,UAAU;AAE5E,SAAS,YAAY,MAA6B;AAChD,QAAM,OAAO,SAAS,MAAM,QAAQ,IAAI,CAAC,EAAE,KAAK;AAEhD,SAAO,SAAS,KAAK,OAAO;AAC9B;","names":[]}

package/dist/source/toc.d.ts

@@ -1,16 +0,0 @@
-import { z } from 'zod';
-
-declare const TOC_FILE_VERSION = 1;
-interface TocItem {
-    readonly title?: string | null | undefined;
-    readonly serialId?: number | undefined;
-    readonly children: readonly TocItem[];
-}
-declare const tocItemSchema: z.ZodType<TocItem>;
-declare const tocFileSchema: z.ZodObject<{
-    version: z.ZodLiteral<1>;
-    items: z.ZodArray<z.ZodType<TocItem, unknown, z.core.$ZodTypeInternals<TocItem, unknown>>>;
-}, z.core.$strip>;
-type TocFile = z.infer<typeof tocFileSchema>;
-
-export { TOC_FILE_VERSION, type TocFile, type TocItem, tocFileSchema, tocItemSchema };

package/dist/source/toc.js

@@ -1,17 +0,0 @@
-import { z } from "zod";
-const TOC_FILE_VERSION = 1;
-const tocItemSchema = z.object({
-  title: z.string().min(1).nullable().optional(),
-  serialId: z.number().int().nonnegative().optional(),
-  children: z.lazy(() => z.array(tocItemSchema))
-});
-const tocFileSchema = z.object({
-  version: z.literal(TOC_FILE_VERSION),
-  items: z.array(tocItemSchema)
-});
-export {
-  TOC_FILE_VERSION,
-  tocFileSchema,
-  tocItemSchema
-};
-//# sourceMappingURL=toc.js.map

package/dist/source/toc.js.map

@@ -1 +0,0 @@
-{"version":3,"sources":["../../src/source/toc.ts"],"sourcesContent":["import { z } from \"zod\";\n\nexport const TOC_FILE_VERSION = 1;\n\nexport interface TocItem {\n  readonly title?: string | null | undefined;\n  readonly serialId?: number | undefined;\n  readonly children: readonly TocItem[];\n}\n\nexport const tocItemSchema: z.ZodType<TocItem> = z.object({\n  title: z.string().min(1).nullable().optional(),\n  serialId: z.number().int().nonnegative().optional(),\n  children: z.lazy(() => z.array(tocItemSchema)),\n});\n\nexport const tocFileSchema = z.object({\n  version: z.literal(TOC_FILE_VERSION),\n  items: z.array(tocItemSchema),\n});\n\nexport type TocFile = z.infer<typeof tocFileSchema>;\n"],"mappings":"AAAA,SAAS,SAAS;AAEX,MAAM,mBAAmB;AAQzB,MAAM,gBAAoC,EAAE,OAAO;AAAA,EACxD,OAAO,EAAE,OAAO,EAAE,IAAI,CAAC,EAAE,SAAS,EAAE,SAAS;AAAA,EAC7C,UAAU,EAAE,OAAO,EAAE,IAAI,EAAE,YAAY,EAAE,SAAS;AAAA,EAClD,UAAU,EAAE,KAAK,MAAM,EAAE,MAAM,aAAa,CAAC;AAC/C,CAAC;AAEM,MAAM,gBAAgB,EAAE,OAAO;AAAA,EACpC,SAAS,EAAE,QAAQ,gBAAgB;AAAA,EACnC,OAAO,EAAE,MAAM,aAAa;AAC9B,CAAC;","names":[]}

package/dist/topology/grouping.d.ts

@@ -1,15 +0,0 @@
-import { SentenceGroupRecord } from '../document/types.js';
-import { computeNormalizedSegmentIncisions } from './segment-incision.js';
-import '../document/text-streams.js';
-import '../document/database.js';
-import 'sqlite3';
-
-declare function groupSegments(input: {
-    edges: Parameters<typeof computeNormalizedSegmentIncisions>[0]["edges"];
-    fragments: Parameters<typeof computeNormalizedSegmentIncisions>[0]["fragments"];
-    groupWordsCount: number;
-    chunks: Parameters<typeof computeNormalizedSegmentIncisions>[0]["chunks"];
-    serialId: number;
-}): Promise<SentenceGroupRecord[]>;
-
-export { groupSegments };

package/dist/topology/grouping.js.map

@@ -1 +0,0 @@
-{"version":3,"sources":["../../src/topology/grouping.ts"],"sourcesContent":["import type { SentenceGroupRecord } from \"../document/index.js\";\nimport { computeNormalizedSegmentIncisions } from \"./segment-incision.js\";\nimport { createSegmentGroups } from \"./resource-segmentation.js\";\n\nexport async function groupSegments(input: {\n  edges: Parameters<typeof computeNormalizedSegmentIncisions>[0][\"edges\"];\n  fragments: Parameters<\n    typeof computeNormalizedSegmentIncisions\n  >[0][\"fragments\"];\n  groupWordsCount: number;\n  chunks: Parameters<typeof computeNormalizedSegmentIncisions>[0][\"chunks\"];\n  serialId: number;\n}): Promise<SentenceGroupRecord[]> {\n  const segmentInfos = await computeNormalizedSegmentIncisions({\n    chunks: input.chunks,\n    edges: input.edges,\n    fragments: input.fragments,\n  });\n\n  return createSegmentGroups({\n    segmentInfos,\n    groupWordsCount: input.groupWordsCount,\n    serialId: input.serialId,\n  });\n}\n"],"mappings":"AACA,SAAS,yCAAyC;AAClD,SAAS,2BAA2B;AAEpC,eAAsB,cAAc,OAQD;AACjC,QAAM,eAAe,MAAM,kCAAkC;AAAA,IAC3D,QAAQ,MAAM;AAAA,IACd,OAAO,MAAM;AAAA,IACb,WAAW,MAAM;AAAA,EACnB,CAAC;AAED,SAAO,oBAAoB;AAAA,IACzB;AAAA,IACA,iBAAiB,MAAM;AAAA,IACvB,UAAU,MAAM;AAAA,EAClB,CAAC;AACH;","names":[]}

package/dist/topology/index.d.ts

@@ -1,21 +0,0 @@
-export { Topology } from './topology.js';
-import '../document/document.js';
-import '../source/meta.js';
-import 'zod';
-import '../source/toc.js';
-import '../source/types.js';
-import '../document/database.js';
-import 'sqlite3';
-import '../document/text-streams.js';
-import '../document/types.js';
-import '../document/stores.js';
-import '../reader/types.js';
-import '../reader/attention/attention.js';
-import '../reader/chunk-batch/types.js';
-import '../common/language.js';
-import '../llm/client.js';
-import '../llm/context.js';
-import '../llm/cache.js';
-import '../llm/types.js';
-import 'ai';
-import '../reader/segment/types.js';

package/dist/topology/index.js

@@ -1,5 +0,0 @@
-import { Topology } from "./topology.js";
-export {
-  Topology
-};
-//# sourceMappingURL=index.js.map

package/dist/topology/index.js.map

@@ -1 +0,0 @@
-{"version":3,"sources":["../../src/topology/index.ts"],"sourcesContent":["export { Topology } from \"./topology.js\";\n"],"mappings":"AAAA,SAAS,gBAAgB;","names":[]}

package/dist/topology/resource-segmentation.d.ts

@@ -1,13 +0,0 @@
-import { SentenceGroupRecord } from '../document/types.js';
-import { SegmentInfo } from './segment-incision.js';
-import '../document/text-streams.js';
-import '../document/database.js';
-import 'sqlite3';
-
-declare function createSegmentGroups(input: {
-    segmentInfos: readonly SegmentInfo[];
-    groupWordsCount: number;
-    serialId: number;
-}): SentenceGroupRecord[];
-
-export { createSegmentGroups };

package/dist/topology/resource-segmentation.js.map

@@ -1 +0,0 @@
-{"version":3,"sources":["../../src/topology/resource-segmentation.ts"],"sourcesContent":["import type { SentenceGroupRecord } from \"../document/index.js\";\nimport type { SegmentInfo } from \"./segment-incision.js\";\n\ninterface SegmentResource {\n  readonly count: number;\n  readonly startIncision: number;\n  readonly endIncision: number;\n  readonly startSentenceIndex: number;\n}\n\nexport function createSegmentGroups(input: {\n  segmentInfos: readonly SegmentInfo[];\n  groupWordsCount: number;\n  serialId: number;\n}): SentenceGroupRecord[] {\n  if (input.segmentInfos.length === 0) {\n    return [];\n  }\n\n  const resources = input.segmentInfos.map(\n    (segmentInfo): SegmentResource => ({\n      count: segmentInfo.wordsCount,\n      endIncision: segmentInfo.endIncision,\n      startSentenceIndex: segmentInfo.startSentenceIndex,\n      startIncision: segmentInfo.startIncision,\n    }),\n  );\n  const groups = allocateSegmentGroups(resources, input.groupWordsCount);\n\n  return groups.map((startSentenceIndexes, groupId) => ({\n    endSentenceIndex: Math.max(...startSentenceIndexes),\n    groupId,\n    serialId: input.serialId,\n    startSentenceIndex: Math.min(...startSentenceIndexes),\n  }));\n}\n\nfunction allocateSegmentGroups(\n  resources: readonly SegmentResource[],\n  maxCount: number,\n): number[][] {\n  return splitResources(resources, maxCount).map((group) =>\n    group.map((resource) => resource.startSentenceIndex),\n  );\n}\n\nfunction splitResources(\n  resources: readonly SegmentResource[],\n  maxCount: number,\n): SegmentResource[][] {\n  if (resources.length === 0) {\n    return [];\n  }\n\n  if (maxCount <= 0) {\n    return resources.map((resource) => [resource]);\n  }\n\n  const totalCount = resources.reduce((total, resource) => {\n    return total + resource.count;\n  }, 0);\n\n  if (resources.length === 1 || totalCount <= maxCount) {\n    return [[...resources]];\n  }\n\n  const splitIndex = findPreferredSplitIndex(resources, maxCount);\n\n  if (splitIndex === undefined) {\n    return splitGreedily(resources, maxCount);\n  }\n\n  const left = resources.slice(0, splitIndex + 1);\n  const right = resources.slice(splitIndex + 1);\n\n  if (\n    left.length === 0 ||\n    right.length === 0 ||\n    left.length === resources.length ||\n    right.length === resources.length\n  ) {\n    return splitGreedily(resources, maxCount);\n  }\n\n  return [\n    ...splitResources(left, maxCount),\n    ...splitResources(right, maxCount),\n  ];\n}\n\nfunction findPreferredSplitIndex(\n  resources: readonly SegmentResource[],\n  maxCount: number,\n): number | undefined {\n  const totalCount = resources.reduce((total, resource) => {\n    return total + resource.count;\n  }, 0);\n  let leftCount = 0;\n  let bestIndex: number | undefined;\n  let bestBoundary = 0;\n  let bestOversizedSides = Number.POSITIVE_INFINITY;\n  let bestOverflow = Number.POSITIVE_INFINITY;\n\n  for (let index = 0; index < resources.length - 1; index += 1) {\n    const current = resources[index];\n    const next = resources[index + 1];\n\n    if (current === undefined || next === undefined) {\n      continue;\n    }\n\n    leftCount += current.count;\n    const rightCount = totalCount - leftCount;\n    const boundary = current.endIncision + next.startIncision;\n\n    if (boundary <= 0) {\n      continue;\n    }\n\n    const oversizedSides =\n      Number(leftCount > maxCount) + Number(rightCount > maxCount);\n    const overflow =\n      Math.max(0, leftCount - maxCount) + Math.max(0, rightCount - maxCount);\n\n    if (\n      oversizedSides < bestOversizedSides ||\n      (oversizedSides === bestOversizedSides && boundary > bestBoundary) ||\n      (oversizedSides === bestOversizedSides &&\n        boundary === bestBoundary &&\n        overflow < bestOverflow)\n    ) {\n      bestBoundary = boundary;\n      bestIndex = index;\n      bestOversizedSides = oversizedSides;\n      bestOverflow = overflow;\n    }\n  }\n\n  return bestIndex;\n}\n\nfunction splitGreedily(\n  resources: readonly SegmentResource[],\n  maxCount: number,\n): SegmentResource[][] {\n  const groups: SegmentResource[][] = [];\n  let currentGroup: SegmentResource[] = [];\n  let currentCount = 0;\n\n  for (const resource of resources) {\n    if (currentGroup.length > 0 && currentCount + resource.count > maxCount) {\n      groups.push(currentGroup);\n      currentGroup = [];\n      currentCount = 0;\n    }\n\n    currentGroup.push(resource);\n    currentCount += resource.count;\n\n    if (currentCount >= maxCount) {\n      groups.push(currentGroup);\n      currentGroup = [];\n      currentCount = 0;\n    }\n  }\n\n  if (currentGroup.length > 0) {\n    groups.push(currentGroup);\n  }\n\n  return groups;\n}\n"],"mappings":"AAUO,SAAS,oBAAoB,OAIV;AACxB,MAAI,MAAM,aAAa,WAAW,GAAG;AACnC,WAAO,CAAC;AAAA,EACV;AAEA,QAAM,YAAY,MAAM,aAAa;AAAA,IACnC,CAAC,iBAAkC;AAAA,MACjC,OAAO,YAAY;AAAA,MACnB,aAAa,YAAY;AAAA,MACzB,oBAAoB,YAAY;AAAA,MAChC,eAAe,YAAY;AAAA,IAC7B;AAAA,EACF;AACA,QAAM,SAAS,sBAAsB,WAAW,MAAM,eAAe;AAErE,SAAO,OAAO,IAAI,CAAC,sBAAsB,aAAa;AAAA,IACpD,kBAAkB,KAAK,IAAI,GAAG,oBAAoB;AAAA,IAClD;AAAA,IACA,UAAU,MAAM;AAAA,IAChB,oBAAoB,KAAK,IAAI,GAAG,oBAAoB;AAAA,EACtD,EAAE;AACJ;AAEA,SAAS,sBACP,WACA,UACY;AACZ,SAAO,eAAe,WAAW,QAAQ,EAAE;AAAA,IAAI,CAAC,UAC9C,MAAM,IAAI,CAAC,aAAa,SAAS,kBAAkB;AAAA,EACrD;AACF;AAEA,SAAS,eACP,WACA,UACqB;AACrB,MAAI,UAAU,WAAW,GAAG;AAC1B,WAAO,CAAC;AAAA,EACV;AAEA,MAAI,YAAY,GAAG;AACjB,WAAO,UAAU,IAAI,CAAC,aAAa,CAAC,QAAQ,CAAC;AAAA,EAC/C;AAEA,QAAM,aAAa,UAAU,OAAO,CAAC,OAAO,aAAa;AACvD,WAAO,QAAQ,SAAS;AAAA,EAC1B,GAAG,CAAC;AAEJ,MAAI,UAAU,WAAW,KAAK,cAAc,UAAU;AACpD,WAAO,CAAC,CAAC,GAAG,SAAS,CAAC;AAAA,EACxB;AAEA,QAAM,aAAa,wBAAwB,WAAW,QAAQ;AAE9D,MAAI,eAAe,QAAW;AAC5B,WAAO,cAAc,WAAW,QAAQ;AAAA,EAC1C;AAEA,QAAM,OAAO,UAAU,MAAM,GAAG,aAAa,CAAC;AAC9C,QAAM,QAAQ,UAAU,MAAM,aAAa,CAAC;AAE5C,MACE,KAAK,WAAW,KAChB,MAAM,WAAW,KACjB,KAAK,WAAW,UAAU,UAC1B,MAAM,WAAW,UAAU,QAC3B;AACA,WAAO,cAAc,WAAW,QAAQ;AAAA,EAC1C;AAEA,SAAO;AAAA,IACL,GAAG,eAAe,MAAM,QAAQ;AAAA,IAChC,GAAG,eAAe,OAAO,QAAQ;AAAA,EACnC;AACF;AAEA,SAAS,wBACP,WACA,UACoB;AACpB,QAAM,aAAa,UAAU,OAAO,CAAC,OAAO,aAAa;AACvD,WAAO,QAAQ,SAAS;AAAA,EAC1B,GAAG,CAAC;AACJ,MAAI,YAAY;AAChB,MAAI;AACJ,MAAI,eAAe;AACnB,MAAI,qBAAqB,OAAO;AAChC,MAAI,eAAe,OAAO;AAE1B,WAAS,QAAQ,GAAG,QAAQ,UAAU,SAAS,GAAG,SAAS,GAAG;AAC5D,UAAM,UAAU,UAAU,KAAK;AAC/B,UAAM,OAAO,UAAU,QAAQ,CAAC;AAEhC,QAAI,YAAY,UAAa,SAAS,QAAW;AAC/C;AAAA,IACF;AAEA,iBAAa,QAAQ;AACrB,UAAM,aAAa,aAAa;AAChC,UAAM,WAAW,QAAQ,cAAc,KAAK;AAE5C,QAAI,YAAY,GAAG;AACjB;AAAA,IACF;AAEA,UAAM,iBACJ,OAAO,YAAY,QAAQ,IAAI,OAAO,aAAa,QAAQ;AAC7D,UAAM,WACJ,KAAK,IAAI,GAAG,YAAY,QAAQ,IAAI,KAAK,IAAI,GAAG,aAAa,QAAQ;AAEvE,QACE,iBAAiB,sBAChB,mBAAmB,sBAAsB,WAAW,gBACpD,mBAAmB,sBAClB,aAAa,gBACb,WAAW,cACb;AACA,qBAAe;AACf,kBAAY;AACZ,2BAAqB;AACrB,qBAAe;AAAA,IACjB;AAAA,EACF;AAEA,SAAO;AACT;AAEA,SAAS,cACP,WACA,UACqB;AACrB,QAAM,SAA8B,CAAC;AACrC,MAAI,eAAkC,CAAC;AACvC,MAAI,eAAe;AAEnB,aAAW,YAAY,WAAW;AAChC,QAAI,aAAa,SAAS,KAAK,eAAe,SAAS,QAAQ,UAAU;AACvE,aAAO,KAAK,YAAY;AACxB,qBAAe,CAAC;AAChB,qBAAe;AAAA,IACjB;AAEA,iBAAa,KAAK,QAAQ;AAC1B,oBAAgB,SAAS;AAEzB,QAAI,gBAAgB,UAAU;AAC5B,aAAO,KAAK,YAAY;AACxB,qBAAe,CAAC;AAChB,qBAAe;AAAA,IACjB;AAAA,EACF;AAEA,MAAI,aAAa,SAAS,GAAG;AAC3B,WAAO,KAAK,YAAY;AAAA,EAC1B;AAEA,SAAO;AACT;","names":[]}

package/dist/topology/segment-incision.d.ts

@@ -1,18 +0,0 @@
-import { ReadonlySerialTextStream } from '../document/text-streams.js';
-import { ChunkRecord, ReadingEdgeRecord } from '../document/types.js';
-import '../document/database.js';
-import 'sqlite3';
-
-interface SegmentInfo {
-    readonly startSentenceIndex: number;
-    readonly wordsCount: number;
-    readonly startIncision: number;
-    readonly endIncision: number;
-}
-declare function computeNormalizedSegmentIncisions(input: {
-    chunks: readonly ChunkRecord[];
-    edges: readonly ReadingEdgeRecord[];
-    fragments: ReadonlySerialTextStream;
-}): Promise<SegmentInfo[]>;
-
-export { type SegmentInfo, computeNormalizedSegmentIncisions };

package/dist/topology/segment-incision.js.map

@@ -1 +0,0 @@
-{"version":3,"sources":["../../src/topology/segment-incision.ts"],"sourcesContent":["import type {\n  ChunkRecord,\n  ReadingEdgeRecord,\n  ReadonlySerialFragments,\n} from \"../document/index.js\";\n\nexport interface SegmentInfo {\n  readonly startSentenceIndex: number;\n  readonly wordsCount: number;\n  readonly startIncision: number;\n  readonly endIncision: number;\n}\n\nexport async function computeNormalizedSegmentIncisions(input: {\n  chunks: readonly ChunkRecord[];\n  edges: readonly ReadingEdgeRecord[];\n  fragments: ReadonlySerialFragments;\n}): Promise<SegmentInfo[]> {\n  const segmentWordsCounts = await loadSegmentWordsCounts(input.fragments);\n\n  return normalizeIncisions(\n    computeSegmentIncisions({\n      chunks: input.chunks,\n      edges: input.edges,\n      segmentWordsCounts,\n    }),\n  );\n}\n\nasync function loadSegmentWordsCounts(\n  fragments: ReadonlySerialFragments,\n): Promise<Readonly<Record<string, number>>> {\n  const wordsCounts = createNumberRecord();\n  const startSentenceIndexes = await fragments.listFragmentIds();\n\n  for (const startSentenceIndex of startSentenceIndexes) {\n    const segment = await fragments.getFragment(startSentenceIndex);\n\n    wordsCounts[String(startSentenceIndex)] = segment.sentences.reduce(\n      (total, sentence) => total + sentence.wordsCount,\n      0,\n    );\n  }\n\n  return wordsCounts;\n}\n\nfunction computeSegmentIncisions(input: {\n  chunks: readonly ChunkRecord[];\n  edges: readonly ReadingEdgeRecord[];\n  segmentWordsCounts: Readonly<Record<string, number>>;\n}): SegmentInfo[] {\n  const startSentenceIndexes = Object.keys(input.segmentWordsCounts)\n    .map(Number)\n    .sort(compareNumber);\n  const chunkIdsBySegmentStartIndex = createNumberListRecord();\n  const chunkWeightsById = createNumberRecord();\n  const edgesByChunkId = createReadingEdgeListRecord();\n\n  for (const chunk of input.chunks) {\n    const startSentenceIndex = findContainingSegmentStartIndex(\n      startSentenceIndexes,\n      chunk.sentenceId[1],\n    );\n\n    if (startSentenceIndex === undefined) {\n      continue;\n    }\n\n    const segmentKey = String(startSentenceIndex);\n\n    chunkWeightsById[String(chunk.id)] = chunk.weight;\n    if (chunkIdsBySegmentStartIndex[segmentKey] === undefined) {\n      chunkIdsBySegmentStartIndex[segmentKey] = [];\n    }\n    chunkIdsBySegmentStartIndex[segmentKey]?.push(chunk.id);\n  }\n\n  for (const edge of input.edges) {\n    const fromKey = String(edge.fromId);\n    const toKey = String(edge.toId);\n\n    if (edgesByChunkId[fromKey] === undefined) {\n      edgesByChunkId[fromKey] = [];\n    }\n    if (edgesByChunkId[toKey] === undefined) {\n      edgesByChunkId[toKey] = [];\n    }\n\n    edgesByChunkId[fromKey]?.push(edge);\n    edgesByChunkId[toKey]?.push(edge);\n  }\n\n  const chunksById = createChunkRecord();\n\n  for (const chunk of input.chunks) {\n    chunksById[String(chunk.id)] = chunk;\n  }\n\n  return startSentenceIndexes\n    .filter(\n      (startSentenceIndex) =>\n        chunkIdsBySegmentStartIndex[String(startSentenceIndex)] !== undefined,\n    )\n    .map((startSentenceIndex) => {\n      const chunkIds =\n        chunkIdsBySegmentStartIndex[String(startSentenceIndex)] ?? [];\n      let endIncision = 0;\n      let startIncision = 0;\n\n      for (const chunkId of chunkIds) {\n        const chunkWeight = chunkWeightsById[String(chunkId)] ?? 0;\n        const chunkEdges = edgesByChunkId[String(chunkId)] ?? [];\n        const externalEdges = chunkEdges\n          .map((edge) => ({\n            otherChunkId: edge.fromId === chunkId ? edge.toId : edge.fromId,\n            weight: edge.weight,\n          }))\n          .map((edge) => {\n            const otherChunk = chunksById[String(edge.otherChunkId)];\n\n            if (otherChunk === undefined) {\n              return undefined;\n            }\n\n            return {\n              otherSegmentStartIndex: findContainingSegmentStartIndex(\n                startSentenceIndexes,\n                otherChunk.sentenceId[1],\n              ),\n              weight: edge.weight,\n            };\n          })\n          .filter(\n            (\n              edge,\n            ): edge is {\n              otherSegmentStartIndex: number;\n              weight: number;\n            } =>\n              edge !== undefined &&\n              edge.otherSegmentStartIndex !== undefined &&\n              edge.otherSegmentStartIndex !== startSentenceIndex,\n          );\n\n        if (externalEdges.length === 0) {\n          continue;\n        }\n\n        const totalExternalWeight = externalEdges.reduce(\n          (total, edge) => total + edge.weight,\n          0,\n        );\n\n        if (totalExternalWeight === 0) {\n          continue;\n        }\n\n        for (const edge of externalEdges) {\n          const halfWeight = (chunkWeight / totalExternalWeight) * edge.weight;\n\n          if (edge.otherSegmentStartIndex < startSentenceIndex) {\n            startIncision += halfWeight;\n            continue;\n          }\n\n          endIncision += halfWeight;\n        }\n      }\n\n      return {\n        endIncision,\n        startSentenceIndex,\n        startIncision,\n        wordsCount: input.segmentWordsCounts[String(startSentenceIndex)] ?? 0,\n      };\n    });\n}\n\nfunction findContainingSegmentStartIndex(\n  startSentenceIndexes: readonly number[],\n  sentenceIndex: number,\n): number | undefined {\n  let containingSegmentStartIndex: number | undefined;\n\n  for (const startSentenceIndex of startSentenceIndexes) {\n    if (startSentenceIndex > sentenceIndex) {\n      break;\n    }\n\n    containingSegmentStartIndex = startSentenceIndex;\n  }\n\n  return containingSegmentStartIndex;\n}\n\nfunction normalizeIncisions(\n  segmentInfos: readonly SegmentInfo[],\n): SegmentInfo[] {\n  const allIncisions = segmentInfos\n    .flatMap((segmentInfo) => [\n      segmentInfo.startIncision,\n      segmentInfo.endIncision,\n    ])\n    .filter((incision) => incision > 0)\n    .sort(compareNumber);\n\n  if (allIncisions.length === 0) {\n    return [...segmentInfos];\n  }\n\n  const thresholdIndex = Math.min(\n    Math.floor(allIncisions.length * 0.98),\n    allIncisions.length - 1,\n  );\n  const threshold = allIncisions[thresholdIndex] ?? 0;\n  const normalValues = allIncisions.filter((incision) => incision < threshold);\n\n  if (normalValues.length === 0) {\n    return segmentInfos.map((segmentInfo) => ({\n      ...segmentInfo,\n      endIncision: segmentInfo.endIncision > 0 ? 10 : 0,\n      startIncision: segmentInfo.startIncision > 0 ? 10 : 0,\n    }));\n  }\n\n  const minValue = Math.min(...normalValues);\n  const maxValue = Math.max(...normalValues);\n  const minLog = minValue > 0 ? Math.log(minValue) : 0;\n  const maxLog = maxValue > 0 ? Math.log(maxValue) : 0;\n  const logRange = maxLog - minLog === 0 ? 1 : maxLog - minLog;\n\n  return segmentInfos.map((segmentInfo) => ({\n    ...segmentInfo,\n    endIncision: normalizeIncision(segmentInfo.endIncision, {\n      logRange,\n      minLog,\n      threshold,\n    }),\n    startIncision: normalizeIncision(segmentInfo.startIncision, {\n      logRange,\n      minLog,\n      threshold,\n    }),\n  }));\n}\n\nfunction normalizeIncision(\n  incision: number,\n  input: {\n    logRange: number;\n    minLog: number;\n    threshold: number;\n  },\n): number {\n  if (incision === 0) {\n    return 0;\n  }\n\n  if (incision >= input.threshold) {\n    return 10;\n  }\n\n  const incisionLog = incision > 0 ? Math.log(incision) : input.minLog;\n\n  return Math.max(\n    1,\n    Math.min(\n      10,\n      Math.round(1 + ((incisionLog - input.minLog) / input.logRange) * 9),\n    ),\n  );\n}\n\nfunction compareNumber(left: number, right: number): number {\n  return left - right;\n}\n\nfunction createChunkRecord(): Record<string, ChunkRecord | undefined> {\n  return Object.create(null) as Record<string, ChunkRecord | undefined>;\n}\n\nfunction createReadingEdgeListRecord(): Record<\n  string,\n  ReadingEdgeRecord[] | undefined\n> {\n  return Object.create(null) as Record<string, ReadingEdgeRecord[] | undefined>;\n}\n\nfunction createNumberListRecord(): Record<string, number[] | undefined> {\n  return Object.create(null) as Record<string, number[] | undefined>;\n}\n\nfunction createNumberRecord(): Record<string, number> {\n  return Object.create(null) as Record<string, number>;\n}\n"],"mappings":"AAaA,eAAsB,kCAAkC,OAI7B;AACzB,QAAM,qBAAqB,MAAM,uBAAuB,MAAM,SAAS;AAEvE,SAAO;AAAA,IACL,wBAAwB;AAAA,MACtB,QAAQ,MAAM;AAAA,MACd,OAAO,MAAM;AAAA,MACb;AAAA,IACF,CAAC;AAAA,EACH;AACF;AAEA,eAAe,uBACb,WAC2C;AAC3C,QAAM,cAAc,mBAAmB;AACvC,QAAM,uBAAuB,MAAM,UAAU,gBAAgB;AAE7D,aAAW,sBAAsB,sBAAsB;AACrD,UAAM,UAAU,MAAM,UAAU,YAAY,kBAAkB;AAE9D,gBAAY,OAAO,kBAAkB,CAAC,IAAI,QAAQ,UAAU;AAAA,MAC1D,CAAC,OAAO,aAAa,QAAQ,SAAS;AAAA,MACtC;AAAA,IACF;AAAA,EACF;AAEA,SAAO;AACT;AAEA,SAAS,wBAAwB,OAIf;AAChB,QAAM,uBAAuB,OAAO,KAAK,MAAM,kBAAkB,EAC9D,IAAI,MAAM,EACV,KAAK,aAAa;AACrB,QAAM,8BAA8B,uBAAuB;AAC3D,QAAM,mBAAmB,mBAAmB;AAC5C,QAAM,iBAAiB,4BAA4B;AAEnD,aAAW,SAAS,MAAM,QAAQ;AAChC,UAAM,qBAAqB;AAAA,MACzB;AAAA,MACA,MAAM,WAAW,CAAC;AAAA,IACpB;AAEA,QAAI,uBAAuB,QAAW;AACpC;AAAA,IACF;AAEA,UAAM,aAAa,OAAO,kBAAkB;AAE5C,qBAAiB,OAAO,MAAM,EAAE,CAAC,IAAI,MAAM;AAC3C,QAAI,4BAA4B,UAAU,MAAM,QAAW;AACzD,kCAA4B,UAAU,IAAI,CAAC;AAAA,IAC7C;AACA,gCAA4B,UAAU,GAAG,KAAK,MAAM,EAAE;AAAA,EACxD;AAEA,aAAW,QAAQ,MAAM,OAAO;AAC9B,UAAM,UAAU,OAAO,KAAK,MAAM;AAClC,UAAM,QAAQ,OAAO,KAAK,IAAI;AAE9B,QAAI,eAAe,OAAO,MAAM,QAAW;AACzC,qBAAe,OAAO,IAAI,CAAC;AAAA,IAC7B;AACA,QAAI,eAAe,KAAK,MAAM,QAAW;AACvC,qBAAe,KAAK,IAAI,CAAC;AAAA,IAC3B;AAEA,mBAAe,OAAO,GAAG,KAAK,IAAI;AAClC,mBAAe,KAAK,GAAG,KAAK,IAAI;AAAA,EAClC;AAEA,QAAM,aAAa,kBAAkB;AAErC,aAAW,SAAS,MAAM,QAAQ;AAChC,eAAW,OAAO,MAAM,EAAE,CAAC,IAAI;AAAA,EACjC;AAEA,SAAO,qBACJ;AAAA,IACC,CAAC,uBACC,4BAA4B,OAAO,kBAAkB,CAAC,MAAM;AAAA,EAChE,EACC,IAAI,CAAC,uBAAuB;AAC3B,UAAM,WACJ,4BAA4B,OAAO,kBAAkB,CAAC,KAAK,CAAC;AAC9D,QAAI,cAAc;AAClB,QAAI,gBAAgB;AAEpB,eAAW,WAAW,UAAU;AAC9B,YAAM,cAAc,iBAAiB,OAAO,OAAO,CAAC,KAAK;AACzD,YAAM,aAAa,eAAe,OAAO,OAAO,CAAC,KAAK,CAAC;AACvD,YAAM,gBAAgB,WACnB,IAAI,CAAC,UAAU;AAAA,QACd,cAAc,KAAK,WAAW,UAAU,KAAK,OAAO,KAAK;AAAA,QACzD,QAAQ,KAAK;AAAA,MACf,EAAE,EACD,IAAI,CAAC,SAAS;AACb,cAAM,aAAa,WAAW,OAAO,KAAK,YAAY,CAAC;AAEvD,YAAI,eAAe,QAAW;AAC5B,iBAAO;AAAA,QACT;AAEA,eAAO;AAAA,UACL,wBAAwB;AAAA,YACtB;AAAA,YACA,WAAW,WAAW,CAAC;AAAA,UACzB;AAAA,UACA,QAAQ,KAAK;AAAA,QACf;AAAA,MACF,CAAC,EACA;AAAA,QACC,CACE,SAKA,SAAS,UACT,KAAK,2BAA2B,UAChC,KAAK,2BAA2B;AAAA,MACpC;AAEF,UAAI,cAAc,WAAW,GAAG;AAC9B;AAAA,MACF;AAEA,YAAM,sBAAsB,cAAc;AAAA,QACxC,CAAC,OAAO,SAAS,QAAQ,KAAK;AAAA,QAC9B;AAAA,MACF;AAEA,UAAI,wBAAwB,GAAG;AAC7B;AAAA,MACF;AAEA,iBAAW,QAAQ,eAAe;AAChC,cAAM,aAAc,cAAc,sBAAuB,KAAK;AAE9D,YAAI,KAAK,yBAAyB,oBAAoB;AACpD,2BAAiB;AACjB;AAAA,QACF;AAEA,uBAAe;AAAA,MACjB;AAAA,IACF;AAEA,WAAO;AAAA,MACL;AAAA,MACA;AAAA,MACA;AAAA,MACA,YAAY,MAAM,mBAAmB,OAAO,kBAAkB,CAAC,KAAK;AAAA,IACtE;AAAA,EACF,CAAC;AACL;AAEA,SAAS,gCACP,sBACA,eACoB;AACpB,MAAI;AAEJ,aAAW,sBAAsB,sBAAsB;AACrD,QAAI,qBAAqB,eAAe;AACtC;AAAA,IACF;AAEA,kCAA8B;AAAA,EAChC;AAEA,SAAO;AACT;AAEA,SAAS,mBACP,cACe;AACf,QAAM,eAAe,aAClB,QAAQ,CAAC,gBAAgB;AAAA,IACxB,YAAY;AAAA,IACZ,YAAY;AAAA,EACd,CAAC,EACA,OAAO,CAAC,aAAa,WAAW,CAAC,EACjC,KAAK,aAAa;AAErB,MAAI,aAAa,WAAW,GAAG;AAC7B,WAAO,CAAC,GAAG,YAAY;AAAA,EACzB;AAEA,QAAM,iBAAiB,KAAK;AAAA,IAC1B,KAAK,MAAM,aAAa,SAAS,IAAI;AAAA,IACrC,aAAa,SAAS;AAAA,EACxB;AACA,QAAM,YAAY,aAAa,cAAc,KAAK;AAClD,QAAM,eAAe,aAAa,OAAO,CAAC,aAAa,WAAW,SAAS;AAE3E,MAAI,aAAa,WAAW,GAAG;AAC7B,WAAO,aAAa,IAAI,CAAC,iBAAiB;AAAA,MACxC,GAAG;AAAA,MACH,aAAa,YAAY,cAAc,IAAI,KAAK;AAAA,MAChD,eAAe,YAAY,gBAAgB,IAAI,KAAK;AAAA,IACtD,EAAE;AAAA,EACJ;AAEA,QAAM,WAAW,KAAK,IAAI,GAAG,YAAY;AACzC,QAAM,WAAW,KAAK,IAAI,GAAG,YAAY;AACzC,QAAM,SAAS,WAAW,IAAI,KAAK,IAAI,QAAQ,IAAI;AACnD,QAAM,SAAS,WAAW,IAAI,KAAK,IAAI,QAAQ,IAAI;AACnD,QAAM,WAAW,SAAS,WAAW,IAAI,IAAI,SAAS;AAEtD,SAAO,aAAa,IAAI,CAAC,iBAAiB;AAAA,IACxC,GAAG;AAAA,IACH,aAAa,kBAAkB,YAAY,aAAa;AAAA,MACtD;AAAA,MACA;AAAA,MACA;AAAA,IACF,CAAC;AAAA,IACD,eAAe,kBAAkB,YAAY,eAAe;AAAA,MAC1D;AAAA,MACA;AAAA,MACA;AAAA,IACF,CAAC;AAAA,EACH,EAAE;AACJ;AAEA,SAAS,kBACP,UACA,OAKQ;AACR,MAAI,aAAa,GAAG;AAClB,WAAO;AAAA,EACT;AAEA,MAAI,YAAY,MAAM,WAAW;AAC/B,WAAO;AAAA,EACT;AAEA,QAAM,cAAc,WAAW,IAAI,KAAK,IAAI,QAAQ,IAAI,MAAM;AAE9D,SAAO,KAAK;AAAA,IACV;AAAA,IACA,KAAK;AAAA,MACH;AAAA,MACA,KAAK,MAAM,KAAM,cAAc,MAAM,UAAU,MAAM,WAAY,CAAC;AAAA,IACpE;AAAA,EACF;AACF;AAEA,SAAS,cAAc,MAAc,OAAuB;AAC1D,SAAO,OAAO;AAChB;AAEA,SAAS,oBAA6D;AACpE,SAAO,uBAAO,OAAO,IAAI;AAC3B;AAEA,SAAS,8BAGP;AACA,SAAO,uBAAO,OAAO,IAAI;AAC3B;AAEA,SAAS,yBAA+D;AACtE,SAAO,uBAAO,OAAO,IAAI;AAC3B;AAEA,SAAS,qBAA6C;AACpD,SAAO,uBAAO,OAAO,IAAI;AAC3B;","names":[]}

package/dist/topology/snake-detector.d.ts

@@ -1,13 +0,0 @@
-import { ChunkRecord, ReadingEdgeRecord } from '../document/types.js';
-
-declare function splitConnectedComponents(input: {
-    chunkIds: readonly number[];
-    edges: readonly ReadingEdgeRecord[];
-}): number[][];
-declare function detectSnakesInComponent(input: {
-    chunks: readonly ChunkRecord[];
-    edges: readonly ReadingEdgeRecord[];
-    snakeWordsCount?: number;
-}): number[][];
-
-export { detectSnakesInComponent, splitConnectedComponents };