wiki-graph-core 0.3.1 → 0.4.0

package/dist/graph/topology/core.js

@@ -0,0 +1,348 @@
+import { groupSegments } from "./grouping.js";
+import {
+  buildSnakeGraph,
+  detectSnakesInComponent,
+  splitConnectedComponents
+} from "./snake/index.js";
+import {
+  computeChunkWeights,
+  computeReadingEdgeWeights,
+  getReadingEdgeKey
+} from "./weights.js";
+const DEFAULT_SNAKE_WORDS_COUNT = 280;
+class Topology {
+  #chunkIds = [];
+  #chunksById = createChunkRecord();
+  #document;
+  #edgeKeys = [];
+  #edgesByKey = createEdgeRecord();
+  #groupWordsCount;
+  #serialId;
+  constructor(document, serialId, groupWordsCount) {
+    this.#document = document;
+    this.#groupWordsCount = groupWordsCount;
+    this.#serialId = serialId;
+  }
+  accept(delta) {
+    for (const chunk of delta.chunks) {
+      this.#saveChunk(chunk);
+    }
+    if (delta.importanceAnnotations !== void 0) {
+      for (const annotation of delta.importanceAnnotations) {
+        const chunk = this.#chunksById[String(annotation.chunkId)];
+        if (chunk === void 0) {
+          continue;
+        }
+        this.#chunksById[String(annotation.chunkId)] = {
+          ...chunk,
+          importance: annotation.importance
+        };
+      }
+    }
+    for (const edge of delta.edges) {
+      this.#saveEdge(edge);
+    }
+  }
+  async finalize() {
+    const chunks = this.#listChunks();
+    const edges = this.#listEdges();
+    const chunkWeights = computeChunkWeights(chunks);
+    const edgeWeights = computeReadingEdgeWeights({
+      chunkWeights,
+      edges
+    });
+    const weightedChunks = chunks.map((chunk) => ({
+      ...chunk,
+      weight: chunkWeights[String(chunk.id)] ?? 0
+    }));
+    const weightedEdges = edges.map((edge) => ({
+      ...edge,
+      weight: edgeWeights[getReadingEdgeKey(edge.fromId, edge.toId)] ?? 0
+    }));
+    const sentenceGroups = await groupSegments({
+      chunks: weightedChunks,
+      edges: weightedEdges,
+      fragments: this.#document.getSerialFragments(this.#serialId),
+      groupWordsCount: this.#groupWordsCount,
+      serialId: this.#serialId
+    });
+    const snakeTopology = buildSnakeTopology({
+      chunks: weightedChunks,
+      edges: weightedEdges,
+      sentenceGroups,
+      serialId: this.#serialId
+    });
+    await this.#document.serials.ensure(this.#serialId);
+    const chunkIdMap = /* @__PURE__ */ new Map();
+    for (const chunk of weightedChunks) {
+      const createdChunk = await this.#document.chunks.create({
+        content: chunk.content,
+        generation: chunk.generation,
+        label: chunk.label,
+        sentenceId: chunk.sentenceId,
+        sentenceIds: chunk.sentenceIds,
+        weight: chunk.weight,
+        wordsCount: chunk.wordsCount,
+        ...chunk.importance === void 0 ? {} : { importance: chunk.importance },
+        ...chunk.retention === void 0 ? {} : { retention: chunk.retention }
+      });
+      chunkIdMap.set(chunk.id, createdChunk.id);
+    }
+    for (const edge of weightedEdges) {
+      const fromId = chunkIdMap.get(edge.fromId);
+      const toId = chunkIdMap.get(edge.toId);
+      if (fromId === void 0 || toId === void 0) {
+        continue;
+      }
+      await this.#document.readingEdges.save({
+        ...edge,
+        fromId,
+        toId
+      });
+    }
+    await this.#document.fragmentGroups.saveMany(sentenceGroups);
+    const snakeIds = [];
+    for (const snake of snakeTopology.snakes) {
+      snakeIds.push(
+        await this.#document.snakes.create({
+          firstLabel: snake.firstLabel,
+          groupId: snake.groupId,
+          lastLabel: snake.lastLabel,
+          localSnakeId: snake.localSnakeId,
+          serialId: this.#serialId,
+          size: snake.size,
+          weight: snake.weight,
+          wordsCount: snake.wordsCount
+        })
+      );
+    }
+    for (const snakeChunk of snakeTopology.snakeChunks) {
+      const snakeId = snakeIds[snakeChunk.snakeIndex];
+      if (snakeId === void 0) {
+        continue;
+      }
+      await this.#document.snakeChunks.save({
+        chunkId: chunkIdMap.get(snakeChunk.chunkId) ?? snakeChunk.chunkId,
+        position: snakeChunk.position,
+        snakeId
+      });
+    }
+    for (const snakeEdge of snakeTopology.snakeEdges) {
+      const fromSnakeId = snakeIds[snakeEdge.fromSnakeIndex];
+      const toSnakeId = snakeIds[snakeEdge.toSnakeIndex];
+      if (fromSnakeId === void 0 || toSnakeId === void 0) {
+        continue;
+      }
+      await this.#document.snakeEdges.save({
+        fromSnakeId,
+        toSnakeId,
+        weight: snakeEdge.weight
+      });
+    }
+  }
+  #listChunks() {
+    return [...this.#chunkIds].sort(compareNumber).map((chunkId) => this.#chunksById[String(chunkId)]).filter((chunk) => chunk !== void 0);
+  }
+  #listEdges() {
+    return [...this.#edgeKeys].sort(compareEdgeKey).map((edgeKey) => this.#edgesByKey[edgeKey]).filter((edge) => edge !== void 0);
+  }
+  #saveChunk(chunk) {
+    const chunkId = String(chunk.id);
+    if (this.#chunksById[chunkId] === void 0) {
+      this.#chunkIds.push(chunk.id);
+    }
+    const importance = chunk.importance;
+    const retention = chunk.retention;
+    this.#chunksById[chunkId] = {
+      content: chunk.content,
+      generation: chunk.generation,
+      id: chunk.id,
+      label: chunk.label,
+      sentenceId: chunk.sentenceId,
+      sentenceIds: [...chunk.sentenceIds],
+      wordsCount: chunk.wordsCount,
+      weight: 0,
+      ...importance === void 0 ? {} : { importance },
+      ...retention === void 0 ? {} : { retention }
+    };
+  }
+  #saveEdge(edge) {
+    const edgeKey = getReadingEdgeKey(edge.fromId, edge.toId);
+    if (this.#edgesByKey[edgeKey] === void 0) {
+      this.#edgeKeys.push(edgeKey);
+    }
+    const strength = edge.strength;
+    this.#edgesByKey[edgeKey] = {
+      fromId: edge.fromId,
+      toId: edge.toId,
+      weight: 0,
+      ...strength === void 0 ? {} : { strength }
+    };
+  }
+}
+function compareEdgeKey(left, right) {
+  const [leftFromIdText = "", leftToIdText = ""] = left.split(":");
+  const [rightFromIdText = "", rightToIdText = ""] = right.split(":");
+  const leftFromId = Number(leftFromIdText);
+  const rightFromId = Number(rightFromIdText);
+  if (leftFromId !== rightFromId) {
+    return leftFromId - rightFromId;
+  }
+  return Number(leftToIdText) - Number(rightToIdText);
+}
+function compareNumber(left, right) {
+  return left - right;
+}
+function createChunkRecord() {
+  return /* @__PURE__ */ Object.create(null);
+}
+function createEdgeRecord() {
+  return /* @__PURE__ */ Object.create(null);
+}
+function buildSnakeTopology(input) {
+  const chunksById = createChunkRecord();
+  const chunkIdsByGroupId = createNumberListRecord();
+  const edgesByGroupId = createReadingEdgeListRecord();
+  const sentenceRangesByGroupId = [];
+  for (const sentenceGroup of input.sentenceGroups) {
+    if (sentenceGroup.serialId !== input.serialId) {
+      continue;
+    }
+    sentenceRangesByGroupId.push(sentenceGroup);
+  }
+  for (const chunk of input.chunks) {
+    const groupId = findSentenceGroupId(
+      sentenceRangesByGroupId,
+      chunk.sentenceId[1]
+    );
+    chunksById[String(chunk.id)] = chunk;
+    if (groupId === void 0) {
+      continue;
+    }
+    if (chunkIdsByGroupId[String(groupId)] === void 0) {
+      chunkIdsByGroupId[String(groupId)] = [];
+    }
+    chunkIdsByGroupId[String(groupId)]?.push(chunk.id);
+  }
+  for (const edge of input.edges) {
+    const fromChunk = chunksById[String(edge.fromId)];
+    const toChunk = chunksById[String(edge.toId)];
+    if (fromChunk === void 0 || toChunk === void 0) {
+      continue;
+    }
+    const fromGroupId = findSentenceGroupId(
+      sentenceRangesByGroupId,
+      fromChunk.sentenceId[1]
+    );
+    const toGroupId = findSentenceGroupId(
+      sentenceRangesByGroupId,
+      toChunk.sentenceId[1]
+    );
+    if (fromGroupId === void 0 || toGroupId === void 0) {
+      continue;
+    }
+    if (fromGroupId !== toGroupId) {
+      continue;
+    }
+    if (edgesByGroupId[String(fromGroupId)] === void 0) {
+      edgesByGroupId[String(fromGroupId)] = [];
+    }
+    edgesByGroupId[String(fromGroupId)]?.push(edge);
+  }
+  const snakes = [];
+  const snakeChunks = [];
+  const snakeEdges = [];
+  for (const groupId of listSortedRecordNumbers(chunkIdsByGroupId)) {
+    const groupChunkIds = [...chunkIdsByGroupId[String(groupId)] ?? []].sort(
+      compareNumber
+    );
+    const groupEdges = edgesByGroupId[String(groupId)] ?? [];
+    const groupSnakeChunkIds = [];
+    for (const componentChunkIds of splitConnectedComponents({
+      chunkIds: groupChunkIds,
+      edges: groupEdges
+    })) {
+      const componentChunkIdRecord = createBooleanRecord();
+      for (const chunkId of componentChunkIds) {
+        componentChunkIdRecord[String(chunkId)] = true;
+      }
+      groupSnakeChunkIds.push(
+        ...detectSnakesInComponent({
+          chunks: componentChunkIds.map((chunkId) => chunksById[String(chunkId)]).filter((chunk) => chunk !== void 0),
+          edges: groupEdges.filter(
+            (edge) => componentChunkIdRecord[String(edge.fromId)] === true && componentChunkIdRecord[String(edge.toId)] === true
+          ),
+          snakeWordsCount: DEFAULT_SNAKE_WORDS_COUNT
+        })
+      );
+    }
+    const snakeStartIndex = snakes.length;
+    for (const [localSnakeId, snakeChunkIds] of groupSnakeChunkIds.entries()) {
+      const firstChunkId = snakeChunkIds[0];
+      const lastChunkId = snakeChunkIds[snakeChunkIds.length - 1];
+      const firstChunk = firstChunkId === void 0 ? void 0 : chunksById[String(firstChunkId)];
+      const lastChunk = lastChunkId === void 0 ? void 0 : chunksById[String(lastChunkId)];
+      if (firstChunk === void 0 || lastChunk === void 0) {
+        continue;
+      }
+      const snakeIndex = snakeStartIndex + localSnakeId;
+      snakes.push({
+        firstLabel: firstChunk.label,
+        groupId,
+        lastLabel: lastChunk.label,
+        localSnakeId,
+        size: snakeChunkIds.length,
+        weight: snakeChunkIds.reduce((sum, chunkId) => {
+          return sum + (chunksById[String(chunkId)]?.weight ?? 0);
+        }, 0),
+        wordsCount: snakeChunkIds.reduce((sum, chunkId) => {
+          return sum + (chunksById[String(chunkId)]?.wordsCount ?? 0);
+        }, 0)
+      });
+      for (const [position, chunkId] of snakeChunkIds.entries()) {
+        snakeChunks.push({
+          chunkId,
+          position,
+          snakeIndex
+        });
+      }
+    }
+    snakeEdges.push(
+      ...buildSnakeGraph({
+        chunksById,
+        edges: groupEdges,
+        snakes: groupSnakeChunkIds
+      }).map((edge) => ({
+        fromSnakeIndex: snakeStartIndex + edge.fromSnakeIndex,
+        toSnakeIndex: snakeStartIndex + edge.toSnakeIndex,
+        weight: edge.weight
+      }))
+    );
+  }
+  return {
+    snakeChunks,
+    snakeEdges,
+    snakes
+  };
+}
+function findSentenceGroupId(groups, sentenceIndex) {
+  return groups.find(
+    (group) => sentenceIndex >= group.startSentenceIndex && sentenceIndex <= group.endSentenceIndex
+  )?.groupId;
+}
+function createBooleanRecord() {
+  return /* @__PURE__ */ Object.create(null);
+}
+function createNumberListRecord() {
+  return /* @__PURE__ */ Object.create(null);
+}
+function createReadingEdgeListRecord() {
+  return /* @__PURE__ */ Object.create(null);
+}
+function listSortedRecordNumbers(record) {
+  return Object.keys(record).map(Number).sort(compareNumber);
+}
+export {
+  Topology
+};
+//# sourceMappingURL=core.js.map

package/dist/graph/topology/core.js.map

@@ -0,0 +1 @@
+{"version":3,"sources":["../../../src/graph/topology/core.ts"],"sourcesContent":["import type {\n  ChunkRecord,\n  Document,\n  ReadingEdgeRecord,\n  SentenceGroupRecord,\n} from \"../../document/index.js\";\nimport type { ReaderChunk, ReaderGraphDelta } from \"../../text/reader/index.js\";\nimport { groupSegments } from \"./grouping.js\";\nimport {\n  buildSnakeGraph,\n  detectSnakesInComponent,\n  splitConnectedComponents,\n} from \"./snake/index.js\";\nimport {\n  computeChunkWeights,\n  computeReadingEdgeWeights,\n  getReadingEdgeKey,\n} from \"./weights.js\";\n\nconst DEFAULT_SNAKE_WORDS_COUNT = 280;\n\nexport class Topology {\n  readonly #chunkIds: number[] = [];\n  readonly #chunksById = createChunkRecord();\n  readonly #document: Document;\n  readonly #edgeKeys: string[] = [];\n  readonly #edgesByKey = createEdgeRecord();\n  readonly #groupWordsCount: number;\n  readonly #serialId: number;\n\n  public constructor(\n    document: Document,\n    serialId: number,\n    groupWordsCount: number,\n  ) {\n    this.#document = document;\n    this.#groupWordsCount = groupWordsCount;\n    this.#serialId = serialId;\n  }\n\n  public accept(delta: ReaderGraphDelta): void {\n    for (const chunk of delta.chunks) {\n      this.#saveChunk(chunk);\n    }\n\n    if (delta.importanceAnnotations !== undefined) {\n      for (const annotation of delta.importanceAnnotations) {\n        const chunk = this.#chunksById[String(annotation.chunkId)];\n\n        if (chunk === undefined) {\n          continue;\n        }\n\n        this.#chunksById[String(annotation.chunkId)] = {\n          ...chunk,\n          importance: annotation.importance,\n        };\n      }\n    }\n\n    for (const edge of delta.edges) {\n      this.#saveEdge(edge);\n    }\n  }\n\n  public async finalize(): Promise<void> {\n    const chunks = this.#listChunks();\n    const edges = this.#listEdges();\n    const chunkWeights = computeChunkWeights(chunks);\n    const edgeWeights = computeReadingEdgeWeights({\n      chunkWeights,\n      edges,\n    });\n    const weightedChunks = chunks.map((chunk) => ({\n      ...chunk,\n      weight: chunkWeights[String(chunk.id)] ?? 0,\n    }));\n    const weightedEdges = edges.map((edge) => ({\n      ...edge,\n      weight: edgeWeights[getReadingEdgeKey(edge.fromId, edge.toId)] ?? 0,\n    }));\n    const sentenceGroups = await groupSegments({\n      chunks: weightedChunks,\n      edges: weightedEdges,\n      fragments: this.#document.getSerialFragments(this.#serialId),\n      groupWordsCount: this.#groupWordsCount,\n      serialId: this.#serialId,\n    });\n    const snakeTopology = buildSnakeTopology({\n      chunks: weightedChunks,\n      edges: weightedEdges,\n      sentenceGroups,\n      serialId: this.#serialId,\n    });\n\n    await this.#document.serials.ensure(this.#serialId);\n\n    const chunkIdMap = new Map<number, number>();\n\n    for (const chunk of weightedChunks) {\n      const createdChunk = await this.#document.chunks.create({\n        content: chunk.content,\n        generation: chunk.generation,\n        label: chunk.label,\n        sentenceId: chunk.sentenceId,\n        sentenceIds: chunk.sentenceIds,\n        weight: chunk.weight,\n        wordsCount: chunk.wordsCount,\n        ...(chunk.importance === undefined\n          ? {}\n          : { importance: chunk.importance }),\n        ...(chunk.retention === undefined\n          ? {}\n          : { retention: chunk.retention }),\n      });\n\n      chunkIdMap.set(chunk.id, createdChunk.id);\n    }\n\n    for (const edge of weightedEdges) {\n      const fromId = chunkIdMap.get(edge.fromId);\n      const toId = chunkIdMap.get(edge.toId);\n\n      if (fromId === undefined || toId === undefined) {\n        continue;\n      }\n\n      await this.#document.readingEdges.save({\n        ...edge,\n        fromId,\n        toId,\n      });\n    }\n\n    await this.#document.fragmentGroups.saveMany(sentenceGroups);\n\n    const snakeIds: number[] = [];\n\n    for (const snake of snakeTopology.snakes) {\n      snakeIds.push(\n        await this.#document.snakes.create({\n          firstLabel: snake.firstLabel,\n          groupId: snake.groupId,\n          lastLabel: snake.lastLabel,\n          localSnakeId: snake.localSnakeId,\n          serialId: this.#serialId,\n          size: snake.size,\n          weight: snake.weight,\n          wordsCount: snake.wordsCount,\n        }),\n      );\n    }\n\n    for (const snakeChunk of snakeTopology.snakeChunks) {\n      const snakeId = snakeIds[snakeChunk.snakeIndex];\n\n      if (snakeId === undefined) {\n        continue;\n      }\n\n      await this.#document.snakeChunks.save({\n        chunkId: chunkIdMap.get(snakeChunk.chunkId) ?? snakeChunk.chunkId,\n        position: snakeChunk.position,\n        snakeId,\n      });\n    }\n\n    for (const snakeEdge of snakeTopology.snakeEdges) {\n      const fromSnakeId = snakeIds[snakeEdge.fromSnakeIndex];\n      const toSnakeId = snakeIds[snakeEdge.toSnakeIndex];\n\n      if (fromSnakeId === undefined || toSnakeId === undefined) {\n        continue;\n      }\n\n      await this.#document.snakeEdges.save({\n        fromSnakeId,\n        toSnakeId,\n        weight: snakeEdge.weight,\n      });\n    }\n  }\n\n  #listChunks(): ChunkRecord[] {\n    return [...this.#chunkIds]\n      .sort(compareNumber)\n      .map((chunkId) => this.#chunksById[String(chunkId)])\n      .filter((chunk): chunk is ChunkRecord => chunk !== undefined);\n  }\n\n  #listEdges(): ReadingEdgeRecord[] {\n    return [...this.#edgeKeys]\n      .sort(compareEdgeKey)\n      .map((edgeKey) => this.#edgesByKey[edgeKey])\n      .filter((edge): edge is ReadingEdgeRecord => edge !== undefined);\n  }\n\n  #saveChunk(chunk: ReaderChunk): void {\n    const chunkId = String(chunk.id);\n\n    if (this.#chunksById[chunkId] === undefined) {\n      this.#chunkIds.push(chunk.id);\n    }\n\n    const importance = chunk.importance;\n    const retention = chunk.retention;\n\n    this.#chunksById[chunkId] = {\n      content: chunk.content,\n      generation: chunk.generation,\n      id: chunk.id,\n      label: chunk.label,\n      sentenceId: chunk.sentenceId,\n      sentenceIds: [...chunk.sentenceIds],\n      wordsCount: chunk.wordsCount,\n      weight: 0,\n      ...(importance === undefined ? {} : { importance }),\n      ...(retention === undefined ? {} : { retention }),\n    };\n  }\n\n  #saveEdge(edge: ReaderGraphDelta[\"edges\"][number]): void {\n    const edgeKey = getReadingEdgeKey(edge.fromId, edge.toId);\n\n    if (this.#edgesByKey[edgeKey] === undefined) {\n      this.#edgeKeys.push(edgeKey);\n    }\n\n    const strength = edge.strength;\n\n    this.#edgesByKey[edgeKey] = {\n      fromId: edge.fromId,\n      toId: edge.toId,\n      weight: 0,\n      ...(strength === undefined ? {} : { strength }),\n    };\n  }\n}\n\nfunction compareEdgeKey(left: string, right: string): number {\n  const [leftFromIdText = \"\", leftToIdText = \"\"] = left.split(\":\");\n  const [rightFromIdText = \"\", rightToIdText = \"\"] = right.split(\":\");\n  const leftFromId = Number(leftFromIdText);\n  const rightFromId = Number(rightFromIdText);\n\n  if (leftFromId !== rightFromId) {\n    return leftFromId - rightFromId;\n  }\n\n  return Number(leftToIdText) - Number(rightToIdText);\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 createEdgeRecord(): Record<string, ReadingEdgeRecord | undefined> {\n  return Object.create(null) as Record<string, ReadingEdgeRecord | undefined>;\n}\n\ninterface SnakeDraft {\n  readonly firstLabel: string;\n  readonly groupId: number;\n  readonly lastLabel: string;\n  readonly localSnakeId: number;\n  readonly size: number;\n  readonly weight: number;\n  readonly wordsCount: number;\n}\n\ninterface SnakeChunkDraft {\n  readonly chunkId: number;\n  readonly position: number;\n  readonly snakeIndex: number;\n}\n\ninterface SnakeEdgeDraft {\n  readonly fromSnakeIndex: number;\n  readonly toSnakeIndex: number;\n  readonly weight: number;\n}\n\nfunction buildSnakeTopology(input: {\n  chunks: readonly ChunkRecord[];\n  edges: readonly ReadingEdgeRecord[];\n  sentenceGroups: readonly SentenceGroupRecord[];\n  serialId: number;\n}): {\n  readonly snakeChunks: readonly SnakeChunkDraft[];\n  readonly snakeEdges: readonly SnakeEdgeDraft[];\n  readonly snakes: readonly SnakeDraft[];\n} {\n  const chunksById = createChunkRecord();\n  const chunkIdsByGroupId = createNumberListRecord();\n  const edgesByGroupId = createReadingEdgeListRecord();\n  const sentenceRangesByGroupId: Array<{\n    readonly endSentenceIndex: number;\n    readonly groupId: number;\n    readonly startSentenceIndex: number;\n  }> = [];\n\n  for (const sentenceGroup of input.sentenceGroups) {\n    if (sentenceGroup.serialId !== input.serialId) {\n      continue;\n    }\n\n    sentenceRangesByGroupId.push(sentenceGroup);\n  }\n\n  for (const chunk of input.chunks) {\n    const groupId = findSentenceGroupId(\n      sentenceRangesByGroupId,\n      chunk.sentenceId[1],\n    );\n\n    chunksById[String(chunk.id)] = chunk;\n    if (groupId === undefined) {\n      continue;\n    }\n    if (chunkIdsByGroupId[String(groupId)] === undefined) {\n      chunkIdsByGroupId[String(groupId)] = [];\n    }\n    chunkIdsByGroupId[String(groupId)]?.push(chunk.id);\n  }\n\n  for (const edge of input.edges) {\n    const fromChunk = chunksById[String(edge.fromId)];\n    const toChunk = chunksById[String(edge.toId)];\n\n    if (fromChunk === undefined || toChunk === undefined) {\n      continue;\n    }\n\n    const fromGroupId = findSentenceGroupId(\n      sentenceRangesByGroupId,\n      fromChunk.sentenceId[1],\n    );\n    const toGroupId = findSentenceGroupId(\n      sentenceRangesByGroupId,\n      toChunk.sentenceId[1],\n    );\n\n    if (fromGroupId === undefined || toGroupId === undefined) {\n      continue;\n    }\n    if (fromGroupId !== toGroupId) {\n      continue;\n    }\n    if (edgesByGroupId[String(fromGroupId)] === undefined) {\n      edgesByGroupId[String(fromGroupId)] = [];\n    }\n\n    edgesByGroupId[String(fromGroupId)]?.push(edge);\n  }\n\n  const snakes: SnakeDraft[] = [];\n  const snakeChunks: SnakeChunkDraft[] = [];\n  const snakeEdges: SnakeEdgeDraft[] = [];\n\n  for (const groupId of listSortedRecordNumbers(chunkIdsByGroupId)) {\n    const groupChunkIds = [...(chunkIdsByGroupId[String(groupId)] ?? [])].sort(\n      compareNumber,\n    );\n    const groupEdges = edgesByGroupId[String(groupId)] ?? [];\n    const groupSnakeChunkIds: number[][] = [];\n\n    for (const componentChunkIds of splitConnectedComponents({\n      chunkIds: groupChunkIds,\n      edges: groupEdges,\n    })) {\n      const componentChunkIdRecord = createBooleanRecord();\n\n      for (const chunkId of componentChunkIds) {\n        componentChunkIdRecord[String(chunkId)] = true;\n      }\n\n      groupSnakeChunkIds.push(\n        ...detectSnakesInComponent({\n          chunks: componentChunkIds\n            .map((chunkId) => chunksById[String(chunkId)])\n            .filter((chunk): chunk is ChunkRecord => chunk !== undefined),\n          edges: groupEdges.filter(\n            (edge) =>\n              componentChunkIdRecord[String(edge.fromId)] === true &&\n              componentChunkIdRecord[String(edge.toId)] === true,\n          ),\n          snakeWordsCount: DEFAULT_SNAKE_WORDS_COUNT,\n        }),\n      );\n    }\n\n    const snakeStartIndex = snakes.length;\n\n    for (const [localSnakeId, snakeChunkIds] of groupSnakeChunkIds.entries()) {\n      const firstChunkId = snakeChunkIds[0];\n      const lastChunkId = snakeChunkIds[snakeChunkIds.length - 1];\n      const firstChunk =\n        firstChunkId === undefined\n          ? undefined\n          : chunksById[String(firstChunkId)];\n      const lastChunk =\n        lastChunkId === undefined ? undefined : chunksById[String(lastChunkId)];\n\n      if (firstChunk === undefined || lastChunk === undefined) {\n        continue;\n      }\n\n      const snakeIndex = snakeStartIndex + localSnakeId;\n\n      snakes.push({\n        firstLabel: firstChunk.label,\n        groupId,\n        lastLabel: lastChunk.label,\n        localSnakeId,\n        size: snakeChunkIds.length,\n        weight: snakeChunkIds.reduce((sum, chunkId) => {\n          return sum + (chunksById[String(chunkId)]?.weight ?? 0);\n        }, 0),\n        wordsCount: snakeChunkIds.reduce((sum, chunkId) => {\n          return sum + (chunksById[String(chunkId)]?.wordsCount ?? 0);\n        }, 0),\n      });\n\n      for (const [position, chunkId] of snakeChunkIds.entries()) {\n        snakeChunks.push({\n          chunkId,\n          position,\n          snakeIndex,\n        });\n      }\n    }\n\n    snakeEdges.push(\n      ...buildSnakeGraph({\n        chunksById,\n        edges: groupEdges,\n        snakes: groupSnakeChunkIds,\n      }).map((edge) => ({\n        fromSnakeIndex: snakeStartIndex + edge.fromSnakeIndex,\n        toSnakeIndex: snakeStartIndex + edge.toSnakeIndex,\n        weight: edge.weight,\n      })),\n    );\n  }\n\n  return {\n    snakeChunks,\n    snakeEdges,\n    snakes,\n  };\n}\n\nfunction findSentenceGroupId(\n  groups: readonly {\n    readonly endSentenceIndex: number;\n    readonly groupId: number;\n    readonly startSentenceIndex: number;\n  }[],\n  sentenceIndex: number,\n): number | undefined {\n  return groups.find(\n    (group) =>\n      sentenceIndex >= group.startSentenceIndex &&\n      sentenceIndex <= group.endSentenceIndex,\n  )?.groupId;\n}\n\nfunction createBooleanRecord(): Record<string, boolean | undefined> {\n  return Object.create(null) as Record<string, boolean | undefined>;\n}\n\nfunction createNumberListRecord(): Record<string, number[] | undefined> {\n  return Object.create(null) as Record<string, number[] | 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 listSortedRecordNumbers(\n  record: Readonly<Record<string, number[] | undefined>>,\n): number[] {\n  return Object.keys(record).map(Number).sort(compareNumber);\n}\n"],"mappings":"AAOA,SAAS,qBAAqB;AAC9B;AAAA,EACE;AAAA,EACA;AAAA,EACA;AAAA,OACK;AACP;AAAA,EACE;AAAA,EACA;AAAA,EACA;AAAA,OACK;AAEP,MAAM,4BAA4B;AAE3B,MAAM,SAAS;AAAA,EACX,YAAsB,CAAC;AAAA,EACvB,cAAc,kBAAkB;AAAA,EAChC;AAAA,EACA,YAAsB,CAAC;AAAA,EACvB,cAAc,iBAAiB;AAAA,EAC/B;AAAA,EACA;AAAA,EAEF,YACL,UACA,UACA,iBACA;AACA,SAAK,YAAY;AACjB,SAAK,mBAAmB;AACxB,SAAK,YAAY;AAAA,EACnB;AAAA,EAEO,OAAO,OAA+B;AAC3C,eAAW,SAAS,MAAM,QAAQ;AAChC,WAAK,WAAW,KAAK;AAAA,IACvB;AAEA,QAAI,MAAM,0BAA0B,QAAW;AAC7C,iBAAW,cAAc,MAAM,uBAAuB;AACpD,cAAM,QAAQ,KAAK,YAAY,OAAO,WAAW,OAAO,CAAC;AAEzD,YAAI,UAAU,QAAW;AACvB;AAAA,QACF;AAEA,aAAK,YAAY,OAAO,WAAW,OAAO,CAAC,IAAI;AAAA,UAC7C,GAAG;AAAA,UACH,YAAY,WAAW;AAAA,QACzB;AAAA,MACF;AAAA,IACF;AAEA,eAAW,QAAQ,MAAM,OAAO;AAC9B,WAAK,UAAU,IAAI;AAAA,IACrB;AAAA,EACF;AAAA,EAEA,MAAa,WAA0B;AACrC,UAAM,SAAS,KAAK,YAAY;AAChC,UAAM,QAAQ,KAAK,WAAW;AAC9B,UAAM,eAAe,oBAAoB,MAAM;AAC/C,UAAM,cAAc,0BAA0B;AAAA,MAC5C;AAAA,MACA;AAAA,IACF,CAAC;AACD,UAAM,iBAAiB,OAAO,IAAI,CAAC,WAAW;AAAA,MAC5C,GAAG;AAAA,MACH,QAAQ,aAAa,OAAO,MAAM,EAAE,CAAC,KAAK;AAAA,IAC5C,EAAE;AACF,UAAM,gBAAgB,MAAM,IAAI,CAAC,UAAU;AAAA,MACzC,GAAG;AAAA,MACH,QAAQ,YAAY,kBAAkB,KAAK,QAAQ,KAAK,IAAI,CAAC,KAAK;AAAA,IACpE,EAAE;AACF,UAAM,iBAAiB,MAAM,cAAc;AAAA,MACzC,QAAQ;AAAA,MACR,OAAO;AAAA,MACP,WAAW,KAAK,UAAU,mBAAmB,KAAK,SAAS;AAAA,MAC3D,iBAAiB,KAAK;AAAA,MACtB,UAAU,KAAK;AAAA,IACjB,CAAC;AACD,UAAM,gBAAgB,mBAAmB;AAAA,MACvC,QAAQ;AAAA,MACR,OAAO;AAAA,MACP;AAAA,MACA,UAAU,KAAK;AAAA,IACjB,CAAC;AAED,UAAM,KAAK,UAAU,QAAQ,OAAO,KAAK,SAAS;AAElD,UAAM,aAAa,oBAAI,IAAoB;AAE3C,eAAW,SAAS,gBAAgB;AAClC,YAAM,eAAe,MAAM,KAAK,UAAU,OAAO,OAAO;AAAA,QACtD,SAAS,MAAM;AAAA,QACf,YAAY,MAAM;AAAA,QAClB,OAAO,MAAM;AAAA,QACb,YAAY,MAAM;AAAA,QAClB,aAAa,MAAM;AAAA,QACnB,QAAQ,MAAM;AAAA,QACd,YAAY,MAAM;AAAA,QAClB,GAAI,MAAM,eAAe,SACrB,CAAC,IACD,EAAE,YAAY,MAAM,WAAW;AAAA,QACnC,GAAI,MAAM,cAAc,SACpB,CAAC,IACD,EAAE,WAAW,MAAM,UAAU;AAAA,MACnC,CAAC;AAED,iBAAW,IAAI,MAAM,IAAI,aAAa,EAAE;AAAA,IAC1C;AAEA,eAAW,QAAQ,eAAe;AAChC,YAAM,SAAS,WAAW,IAAI,KAAK,MAAM;AACzC,YAAM,OAAO,WAAW,IAAI,KAAK,IAAI;AAErC,UAAI,WAAW,UAAa,SAAS,QAAW;AAC9C;AAAA,MACF;AAEA,YAAM,KAAK,UAAU,aAAa,KAAK;AAAA,QACrC,GAAG;AAAA,QACH;AAAA,QACA;AAAA,MACF,CAAC;AAAA,IACH;AAEA,UAAM,KAAK,UAAU,eAAe,SAAS,cAAc;AAE3D,UAAM,WAAqB,CAAC;AAE5B,eAAW,SAAS,cAAc,QAAQ;AACxC,eAAS;AAAA,QACP,MAAM,KAAK,UAAU,OAAO,OAAO;AAAA,UACjC,YAAY,MAAM;AAAA,UAClB,SAAS,MAAM;AAAA,UACf,WAAW,MAAM;AAAA,UACjB,cAAc,MAAM;AAAA,UACpB,UAAU,KAAK;AAAA,UACf,MAAM,MAAM;AAAA,UACZ,QAAQ,MAAM;AAAA,UACd,YAAY,MAAM;AAAA,QACpB,CAAC;AAAA,MACH;AAAA,IACF;AAEA,eAAW,cAAc,cAAc,aAAa;AAClD,YAAM,UAAU,SAAS,WAAW,UAAU;AAE9C,UAAI,YAAY,QAAW;AACzB;AAAA,MACF;AAEA,YAAM,KAAK,UAAU,YAAY,KAAK;AAAA,QACpC,SAAS,WAAW,IAAI,WAAW,OAAO,KAAK,WAAW;AAAA,QAC1D,UAAU,WAAW;AAAA,QACrB;AAAA,MACF,CAAC;AAAA,IACH;AAEA,eAAW,aAAa,cAAc,YAAY;AAChD,YAAM,cAAc,SAAS,UAAU,cAAc;AACrD,YAAM,YAAY,SAAS,UAAU,YAAY;AAEjD,UAAI,gBAAgB,UAAa,cAAc,QAAW;AACxD;AAAA,MACF;AAEA,YAAM,KAAK,UAAU,WAAW,KAAK;AAAA,QACnC;AAAA,QACA;AAAA,QACA,QAAQ,UAAU;AAAA,MACpB,CAAC;AAAA,IACH;AAAA,EACF;AAAA,EAEA,cAA6B;AAC3B,WAAO,CAAC,GAAG,KAAK,SAAS,EACtB,KAAK,aAAa,EAClB,IAAI,CAAC,YAAY,KAAK,YAAY,OAAO,OAAO,CAAC,CAAC,EAClD,OAAO,CAAC,UAAgC,UAAU,MAAS;AAAA,EAChE;AAAA,EAEA,aAAkC;AAChC,WAAO,CAAC,GAAG,KAAK,SAAS,EACtB,KAAK,cAAc,EACnB,IAAI,CAAC,YAAY,KAAK,YAAY,OAAO,CAAC,EAC1C,OAAO,CAAC,SAAoC,SAAS,MAAS;AAAA,EACnE;AAAA,EAEA,WAAW,OAA0B;AACnC,UAAM,UAAU,OAAO,MAAM,EAAE;AAE/B,QAAI,KAAK,YAAY,OAAO,MAAM,QAAW;AAC3C,WAAK,UAAU,KAAK,MAAM,EAAE;AAAA,IAC9B;AAEA,UAAM,aAAa,MAAM;AACzB,UAAM,YAAY,MAAM;AAExB,SAAK,YAAY,OAAO,IAAI;AAAA,MAC1B,SAAS,MAAM;AAAA,MACf,YAAY,MAAM;AAAA,MAClB,IAAI,MAAM;AAAA,MACV,OAAO,MAAM;AAAA,MACb,YAAY,MAAM;AAAA,MAClB,aAAa,CAAC,GAAG,MAAM,WAAW;AAAA,MAClC,YAAY,MAAM;AAAA,MAClB,QAAQ;AAAA,MACR,GAAI,eAAe,SAAY,CAAC,IAAI,EAAE,WAAW;AAAA,MACjD,GAAI,cAAc,SAAY,CAAC,IAAI,EAAE,UAAU;AAAA,IACjD;AAAA,EACF;AAAA,EAEA,UAAU,MAA+C;AACvD,UAAM,UAAU,kBAAkB,KAAK,QAAQ,KAAK,IAAI;AAExD,QAAI,KAAK,YAAY,OAAO,MAAM,QAAW;AAC3C,WAAK,UAAU,KAAK,OAAO;AAAA,IAC7B;AAEA,UAAM,WAAW,KAAK;AAEtB,SAAK,YAAY,OAAO,IAAI;AAAA,MAC1B,QAAQ,KAAK;AAAA,MACb,MAAM,KAAK;AAAA,MACX,QAAQ;AAAA,MACR,GAAI,aAAa,SAAY,CAAC,IAAI,EAAE,SAAS;AAAA,IAC/C;AAAA,EACF;AACF;AAEA,SAAS,eAAe,MAAc,OAAuB;AAC3D,QAAM,CAAC,iBAAiB,IAAI,eAAe,EAAE,IAAI,KAAK,MAAM,GAAG;AAC/D,QAAM,CAAC,kBAAkB,IAAI,gBAAgB,EAAE,IAAI,MAAM,MAAM,GAAG;AAClE,QAAM,aAAa,OAAO,cAAc;AACxC,QAAM,cAAc,OAAO,eAAe;AAE1C,MAAI,eAAe,aAAa;AAC9B,WAAO,aAAa;AAAA,EACtB;AAEA,SAAO,OAAO,YAAY,IAAI,OAAO,aAAa;AACpD;AAEA,SAAS,cAAc,MAAc,OAAuB;AAC1D,SAAO,OAAO;AAChB;AAEA,SAAS,oBAA6D;AACpE,SAAO,uBAAO,OAAO,IAAI;AAC3B;AAEA,SAAS,mBAAkE;AACzE,SAAO,uBAAO,OAAO,IAAI;AAC3B;AAwBA,SAAS,mBAAmB,OAS1B;AACA,QAAM,aAAa,kBAAkB;AACrC,QAAM,oBAAoB,uBAAuB;AACjD,QAAM,iBAAiB,4BAA4B;AACnD,QAAM,0BAID,CAAC;AAEN,aAAW,iBAAiB,MAAM,gBAAgB;AAChD,QAAI,cAAc,aAAa,MAAM,UAAU;AAC7C;AAAA,IACF;AAEA,4BAAwB,KAAK,aAAa;AAAA,EAC5C;AAEA,aAAW,SAAS,MAAM,QAAQ;AAChC,UAAM,UAAU;AAAA,MACd;AAAA,MACA,MAAM,WAAW,CAAC;AAAA,IACpB;AAEA,eAAW,OAAO,MAAM,EAAE,CAAC,IAAI;AAC/B,QAAI,YAAY,QAAW;AACzB;AAAA,IACF;AACA,QAAI,kBAAkB,OAAO,OAAO,CAAC,MAAM,QAAW;AACpD,wBAAkB,OAAO,OAAO,CAAC,IAAI,CAAC;AAAA,IACxC;AACA,sBAAkB,OAAO,OAAO,CAAC,GAAG,KAAK,MAAM,EAAE;AAAA,EACnD;AAEA,aAAW,QAAQ,MAAM,OAAO;AAC9B,UAAM,YAAY,WAAW,OAAO,KAAK,MAAM,CAAC;AAChD,UAAM,UAAU,WAAW,OAAO,KAAK,IAAI,CAAC;AAE5C,QAAI,cAAc,UAAa,YAAY,QAAW;AACpD;AAAA,IACF;AAEA,UAAM,cAAc;AAAA,MAClB;AAAA,MACA,UAAU,WAAW,CAAC;AAAA,IACxB;AACA,UAAM,YAAY;AAAA,MAChB;AAAA,MACA,QAAQ,WAAW,CAAC;AAAA,IACtB;AAEA,QAAI,gBAAgB,UAAa,cAAc,QAAW;AACxD;AAAA,IACF;AACA,QAAI,gBAAgB,WAAW;AAC7B;AAAA,IACF;AACA,QAAI,eAAe,OAAO,WAAW,CAAC,MAAM,QAAW;AACrD,qBAAe,OAAO,WAAW,CAAC,IAAI,CAAC;AAAA,IACzC;AAEA,mBAAe,OAAO,WAAW,CAAC,GAAG,KAAK,IAAI;AAAA,EAChD;AAEA,QAAM,SAAuB,CAAC;AAC9B,QAAM,cAAiC,CAAC;AACxC,QAAM,aAA+B,CAAC;AAEtC,aAAW,WAAW,wBAAwB,iBAAiB,GAAG;AAChE,UAAM,gBAAgB,CAAC,GAAI,kBAAkB,OAAO,OAAO,CAAC,KAAK,CAAC,CAAE,EAAE;AAAA,MACpE;AAAA,IACF;AACA,UAAM,aAAa,eAAe,OAAO,OAAO,CAAC,KAAK,CAAC;AACvD,UAAM,qBAAiC,CAAC;AAExC,eAAW,qBAAqB,yBAAyB;AAAA,MACvD,UAAU;AAAA,MACV,OAAO;AAAA,IACT,CAAC,GAAG;AACF,YAAM,yBAAyB,oBAAoB;AAEnD,iBAAW,WAAW,mBAAmB;AACvC,+BAAuB,OAAO,OAAO,CAAC,IAAI;AAAA,MAC5C;AAEA,yBAAmB;AAAA,QACjB,GAAG,wBAAwB;AAAA,UACzB,QAAQ,kBACL,IAAI,CAAC,YAAY,WAAW,OAAO,OAAO,CAAC,CAAC,EAC5C,OAAO,CAAC,UAAgC,UAAU,MAAS;AAAA,UAC9D,OAAO,WAAW;AAAA,YAChB,CAAC,SACC,uBAAuB,OAAO,KAAK,MAAM,CAAC,MAAM,QAChD,uBAAuB,OAAO,KAAK,IAAI,CAAC,MAAM;AAAA,UAClD;AAAA,UACA,iBAAiB;AAAA,QACnB,CAAC;AAAA,MACH;AAAA,IACF;AAEA,UAAM,kBAAkB,OAAO;AAE/B,eAAW,CAAC,cAAc,aAAa,KAAK,mBAAmB,QAAQ,GAAG;AACxE,YAAM,eAAe,cAAc,CAAC;AACpC,YAAM,cAAc,cAAc,cAAc,SAAS,CAAC;AAC1D,YAAM,aACJ,iBAAiB,SACb,SACA,WAAW,OAAO,YAAY,CAAC;AACrC,YAAM,YACJ,gBAAgB,SAAY,SAAY,WAAW,OAAO,WAAW,CAAC;AAExE,UAAI,eAAe,UAAa,cAAc,QAAW;AACvD;AAAA,MACF;AAEA,YAAM,aAAa,kBAAkB;AAErC,aAAO,KAAK;AAAA,QACV,YAAY,WAAW;AAAA,QACvB;AAAA,QACA,WAAW,UAAU;AAAA,QACrB;AAAA,QACA,MAAM,cAAc;AAAA,QACpB,QAAQ,cAAc,OAAO,CAAC,KAAK,YAAY;AAC7C,iBAAO,OAAO,WAAW,OAAO,OAAO,CAAC,GAAG,UAAU;AAAA,QACvD,GAAG,CAAC;AAAA,QACJ,YAAY,cAAc,OAAO,CAAC,KAAK,YAAY;AACjD,iBAAO,OAAO,WAAW,OAAO,OAAO,CAAC,GAAG,cAAc;AAAA,QAC3D,GAAG,CAAC;AAAA,MACN,CAAC;AAED,iBAAW,CAAC,UAAU,OAAO,KAAK,cAAc,QAAQ,GAAG;AACzD,oBAAY,KAAK;AAAA,UACf;AAAA,UACA;AAAA,UACA;AAAA,QACF,CAAC;AAAA,MACH;AAAA,IACF;AAEA,eAAW;AAAA,MACT,GAAG,gBAAgB;AAAA,QACjB;AAAA,QACA,OAAO;AAAA,QACP,QAAQ;AAAA,MACV,CAAC,EAAE,IAAI,CAAC,UAAU;AAAA,QAChB,gBAAgB,kBAAkB,KAAK;AAAA,QACvC,cAAc,kBAAkB,KAAK;AAAA,QACrC,QAAQ,KAAK;AAAA,MACf,EAAE;AAAA,IACJ;AAAA,EACF;AAEA,SAAO;AAAA,IACL;AAAA,IACA;AAAA,IACA;AAAA,EACF;AACF;AAEA,SAAS,oBACP,QAKA,eACoB;AACpB,SAAO,OAAO;AAAA,IACZ,CAAC,UACC,iBAAiB,MAAM,sBACvB,iBAAiB,MAAM;AAAA,EAC3B,GAAG;AACL;AAEA,SAAS,sBAA2D;AAClE,SAAO,uBAAO,OAAO,IAAI;AAC3B;AAEA,SAAS,yBAA+D;AACtE,SAAO,uBAAO,OAAO,IAAI;AAC3B;AAEA,SAAS,8BAGP;AACA,SAAO,uBAAO,OAAO,IAAI;AAC3B;AAEA,SAAS,wBACP,QACU;AACV,SAAO,OAAO,KAAK,MAAM,EAAE,IAAI,MAAM,EAAE,KAAK,aAAa;AAC3D;","names":[]}

package/dist/graph/topology/grouping.d.ts

@@ -0,0 +1,13 @@
+import { SentenceGroupRecord } from '../../document/types.js';
+import { computeNormalizedSegmentIncisions } from './segment-incision.js';
+import '../../document/text-streams/types.js';
+
+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/graph/topology/grouping.js.map

@@ -0,0 +1 @@
+{"version":3,"sources":["../../../src/graph/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/graph/topology/index.d.ts

@@ -0,0 +1,34 @@
+export { Topology } from './core.js';
+import '../../document/directory/types.js';
+import '../../text/source/meta.js';
+import 'zod';
+import '../../text/source/toc.js';
+import '../../text/source/types.js';
+import '../../document/database.js';
+import 'sqlite3';
+import '../../document/types.js';
+import '../../document/text-streams/serial.js';
+import '../../document/text-streams/draft.js';
+import '../../document/text-streams/types.js';
+import '../../document/stores/chunk.js';
+import '../../document/stores/types.js';
+import '../../document/stores/fragment-group.js';
+import '../../document/stores/graph-build-parameter.js';
+import '../../document/stores/mention-link.js';
+import '../../document/stores/mention.js';
+import '../../document/stores/object-metadata.js';
+import '../../document/stores/reading-edge.js';
+import '../../document/stores/serial.js';
+import '../../document/stores/snake/chunk.js';
+import '../../document/stores/snake/core.js';
+import '../../document/stores/snake/edge.js';
+import '../../text/reader/types.js';
+import '../../text/reader/attention/core.js';
+import '../../text/reader/chunk-batch/types.js';
+import '../../runtime/common/language.js';
+import '../../external/llm/client/core.js';
+import '../../external/llm/context.js';
+import '../../external/llm/cache.js';
+import '../../external/llm/types.js';
+import 'ai';
+import '../../text/reader/segment/types.js';

package/dist/graph/topology/index.js

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

package/dist/graph/topology/index.js.map

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

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

@@ -0,0 +1,11 @@
+import { SentenceGroupRecord } from '../../document/types.js';
+import { SegmentInfo } from './segment-incision.js';
+import '../../document/text-streams/types.js';
+
+declare function createSegmentGroups(input: {
+    segmentInfos: readonly SegmentInfo[];
+    groupWordsCount: number;
+    serialId: number;
+}): SentenceGroupRecord[];
+
+export { createSegmentGroups };

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

@@ -0,0 +1 @@
+{"version":3,"sources":["../../../src/graph/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/graph/topology/segment-incision.d.ts

@@ -0,0 +1,16 @@
+import { ChunkRecord, ReadingEdgeRecord } from '../../document/types.js';
+import { ReadonlySerialTextStream } from '../../document/text-streams/types.js';
+
+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/graph/topology/segment-incision.js.map

@@ -0,0 +1 @@
+{"version":3,"sources":["../../../src/graph/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/graph/topology/snake/compare.d.ts

@@ -0,0 +1,8 @@
+import { ChunkRecord } from '../../../document/types.js';
+import { EdgeQueueEntry } from './types.js';
+
+declare function compareChunkBySentence(left: ChunkRecord, right: ChunkRecord): number;
+declare function compareEdgeQueueEntry(left: EdgeQueueEntry, right: EdgeQueueEntry): number;
+declare function compareNumber(left: number, right: number): number;
+
+export { compareChunkBySentence, compareEdgeQueueEntry, compareNumber };

package/dist/graph/topology/snake/compare.js

@@ -0,0 +1,29 @@
+function compareChunkBySentence(left, right) {
+  const [leftSerialId, leftSentenceIndex] = left.sentenceId;
+  const [rightSerialId, rightSentenceIndex] = right.sentenceId;
+  if (leftSerialId !== rightSerialId) {
+    return leftSerialId - rightSerialId;
+  }
+  if (leftSentenceIndex !== rightSentenceIndex) {
+    return leftSentenceIndex - rightSentenceIndex;
+  }
+  return left.id - right.id;
+}
+function compareEdgeQueueEntry(left, right) {
+  if (left.value !== right.value) {
+    return right.value - left.value;
+  }
+  if (left.leftClusterId !== right.leftClusterId) {
+    return left.leftClusterId - right.leftClusterId;
+  }
+  return left.rightClusterId - right.rightClusterId;
+}
+function compareNumber(left, right) {
+  return left - right;
+}
+export {
+  compareChunkBySentence,
+  compareEdgeQueueEntry,
+  compareNumber
+};
+//# sourceMappingURL=compare.js.map

package/dist/graph/topology/snake/compare.js.map

@@ -0,0 +1 @@
+{"version":3,"sources":["../../../../src/graph/topology/snake/compare.ts"],"sourcesContent":["import type { ChunkRecord } from \"../../../document/index.js\";\nimport type { EdgeQueueEntry } from \"./types.js\";\n\nexport function compareChunkBySentence(\n  left: ChunkRecord,\n  right: ChunkRecord,\n): number {\n  const [leftSerialId, leftSentenceIndex] = left.sentenceId;\n  const [rightSerialId, rightSentenceIndex] = right.sentenceId;\n\n  if (leftSerialId !== rightSerialId) {\n    return leftSerialId - rightSerialId;\n  }\n\n  if (leftSentenceIndex !== rightSentenceIndex) {\n    return leftSentenceIndex - rightSentenceIndex;\n  }\n\n  return left.id - right.id;\n}\n\nexport function compareEdgeQueueEntry(\n  left: EdgeQueueEntry,\n  right: EdgeQueueEntry,\n): number {\n  if (left.value !== right.value) {\n    return right.value - left.value;\n  }\n\n  if (left.leftClusterId !== right.leftClusterId) {\n    return left.leftClusterId - right.leftClusterId;\n  }\n\n  return left.rightClusterId - right.rightClusterId;\n}\n\nexport function compareNumber(left: number, right: number): number {\n  return left - right;\n}\n"],"mappings":"AAGO,SAAS,uBACd,MACA,OACQ;AACR,QAAM,CAAC,cAAc,iBAAiB,IAAI,KAAK;AAC/C,QAAM,CAAC,eAAe,kBAAkB,IAAI,MAAM;AAElD,MAAI,iBAAiB,eAAe;AAClC,WAAO,eAAe;AAAA,EACxB;AAEA,MAAI,sBAAsB,oBAAoB;AAC5C,WAAO,oBAAoB;AAAA,EAC7B;AAEA,SAAO,KAAK,KAAK,MAAM;AACzB;AAEO,SAAS,sBACd,MACA,OACQ;AACR,MAAI,KAAK,UAAU,MAAM,OAAO;AAC9B,WAAO,MAAM,QAAQ,KAAK;AAAA,EAC5B;AAEA,MAAI,KAAK,kBAAkB,MAAM,eAAe;AAC9C,WAAO,KAAK,gBAAgB,MAAM;AAAA,EACpC;AAEA,SAAO,KAAK,iBAAiB,MAAM;AACrC;AAEO,SAAS,cAAc,MAAc,OAAuB;AACjE,SAAO,OAAO;AAChB;","names":[]}

package/dist/graph/topology/snake/components.d.ts

@@ -0,0 +1,12 @@
+import { ReadingEdgeRecord } from '../../../document/types.js';
+
+declare function splitConnectedComponents(input: {
+    chunkIds: readonly number[];
+    edges: readonly ReadingEdgeRecord[];
+}): number[][];
+declare function isWeaklyConnected(graph: {
+    readonly sortedChunkIds: readonly number[];
+    readonly undirectedAdjacentChunkIdsByChunkId: Readonly<Record<string, readonly number[] | undefined>>;
+}): boolean;
+
+export { isWeaklyConnected, splitConnectedComponents };