update at 2026-02-12 17:30:41

src/core/index.ts

@@ -0,0 +1,3 @@
+export * from './types';
+export * from './parser';
+export * from './sankey';

src/core/parser.ts

@@ -0,0 +1,96 @@
+import Papa from 'papaparse';
+import * as XLSX from 'xlsx';
+import type { RawTable } from './types';
+
+/**
+ * 将任意单元格值转换为字符串，统一处理 null/undefined 场景。
+ */
+function normalizeCell(value: unknown): string {
+  if (value === null || value === undefined) {
+    return '';
+  }
+  return String(value).trim();
+}
+
+/**
+ * 将二维数组标准化为 RawTable。
+ * 约定第一行为表头，后续为数据行。
+ */
+function toRawTable(rows: unknown[][]): RawTable {
+  if (rows.length === 0) {
+    return { headers: [], rows: [] };
+  }
+
+  const firstRow = rows[0] ?? [];
+  const maxColumns = rows.reduce((max, row) => Math.max(max, row.length), firstRow.length);
+
+  const headers = Array.from({ length: maxColumns }, (_, index) => {
+    const header = normalizeCell(firstRow[index]);
+    return header || `列${index + 1}`;
+  });
+
+  const dataRows = rows.slice(1).map((row) => {
+    return Array.from({ length: maxColumns }, (_, index) => normalizeCell(row[index]));
+  });
+
+  return {
+    headers,
+    rows: dataRows
+  };
+}
+
+/**
+ * 解析 CSV 文本为统一表结构。
+ */
+export function parseCsvText(csvText: string): RawTable {
+  const parsed = Papa.parse<string[]>(csvText, {
+    skipEmptyLines: false
+  });
+
+  if (parsed.errors.length > 0) {
+    const firstError = parsed.errors[0];
+    throw new Error(`CSV 解析失败: ${firstError.message}`);
+  }
+
+  const rows = parsed.data.map((row: string[]) => row ?? []);
+  return toRawTable(rows);
+}
+
+/**
+ * 解析 xlsx 的二进制数据。
+ */
+export function parseXlsxBuffer(buffer: ArrayBuffer): RawTable {
+  const workbook = XLSX.read(buffer, { type: 'array' });
+  const firstSheetName = workbook.SheetNames[0];
+
+  if (!firstSheetName) {
+    throw new Error('Excel 文件中没有工作表');
+  }
+
+  const sheet = workbook.Sheets[firstSheetName];
+  const rows = XLSX.utils.sheet_to_json<unknown[]>(sheet, {
+    header: 1,
+    raw: false,
+    defval: ''
+  });
+
+  return toRawTable(rows);
+}
+
+/**
+ * 根据文件后缀自动判断并解析文件。
+ */
+export async function parseDataFile(file: File): Promise<RawTable> {
+  const lowerName = file.name.toLowerCase();
+  if (lowerName.endsWith('.csv')) {
+    const text = await file.text();
+    return parseCsvText(text);
+  }
+
+  if (lowerName.endsWith('.xlsx') || lowerName.endsWith('.xls')) {
+    const buffer = await file.arrayBuffer();
+    return parseXlsxBuffer(buffer);
+  }
+
+  throw new Error('仅支持 .csv / .xlsx / .xls 文件');
+}

src/core/sankey.ts

@@ -0,0 +1,168 @@
+import type {
+  DirectionMode,
+  MappingConfig,
+  RawTable,
+  SankeyBuildResult,
+  SankeyLink,
+  SankeyNode
+} from './types';
+
+/**
+ * 统一清洗字符串，避免因为前后空格导致节点重复。
+ */
+function normalizeText(value: string): string {
+  return value.trim();
+}
+
+/**
+ * 将字符串解析为数字，支持千分位（例如 12,000）。
+ */
+function parseNumericValue(text: string): number | null {
+  const normalized = text.replace(/,/g, '').trim();
+  if (!normalized) {
+    return null;
+  }
+
+  const parsed = Number(normalized);
+  if (Number.isNaN(parsed)) {
+    return null;
+  }
+
+  return parsed;
+}
+
+/**
+ * 按照配置生成 source 名称。
+ * 若未选择描述列，则回退为数据列文本。
+ */
+function buildSourceName(row: string[], config: MappingConfig): string {
+  const sourceDataValue = config.sourceDataColumn === null ? '' : row[config.sourceDataColumn] ?? '';
+
+  if (config.sourceDescriptionColumns.length === 0) {
+    return normalizeText(sourceDataValue);
+  }
+
+  const parts = config.sourceDescriptionColumns
+    .map((column) => normalizeText(row[column] ?? ''))
+    .filter((item) => item.length > 0);
+
+  return parts.join(config.delimiter);
+}
+
+/**
+ * 生成 target 名称，并实现“合并单元格向下补全”的语义。
+ */
+function buildTargetName(
+  row: string[],
+  config: MappingConfig,
+  lastNonEmptyTargetValueByColumn: Map<number, string>
+): string {
+  const parts = config.targetDescriptionColumns
+    .map((column) => {
+      const raw = normalizeText(row[column] ?? '');
+      if (raw.length > 0) {
+        lastNonEmptyTargetValueByColumn.set(column, raw);
+        return raw;
+      }
+
+      return lastNonEmptyTargetValueByColumn.get(column) ?? '';
+    })
+    .filter((item) => item.length > 0);
+
+  return parts.join(config.delimiter);
+}
+
+/**
+ * 将映射配置应用到表格数据，输出桑基图节点和连线。
+ */
+export function buildSankeyData(table: RawTable, config: MappingConfig): SankeyBuildResult {
+  if (config.sourceDataColumn === null) {
+    throw new Error('必须选择源数据列');
+  }
+
+  if (config.targetDescriptionColumns.length === 0) {
+    throw new Error('必须至少选择一个目标描述列');
+  }
+
+  const linkValueMap = new Map<string, number>();
+  const warnings: string[] = [];
+  let droppedRows = 0;
+
+  const lastNonEmptyTargetValueByColumn = new Map<number, string>();
+
+  table.rows.forEach((row, rowIndex) => {
+    const excelRow = rowIndex + 2;
+    const sourceRaw = normalizeText(row[config.sourceDataColumn as number] ?? '');
+    const sourceValue = parseNumericValue(sourceRaw);
+
+    if (sourceValue === null) {
+      warnings.push(`第 ${excelRow} 行: 源数据不是有效数字，已跳过`);
+      droppedRows += 1;
+      return;
+    }
+
+    const sourceName = buildSourceName(row, config);
+    if (!sourceName) {
+      warnings.push(`第 ${excelRow} 行: 源描述为空，已跳过`);
+      droppedRows += 1;
+      return;
+    }
+
+    const targetName = buildTargetName(row, config, lastNonEmptyTargetValueByColumn);
+    if (!targetName) {
+      warnings.push(`第 ${excelRow} 行: 目标描述为空，已跳过`);
+      droppedRows += 1;
+      return;
+    }
+
+    const key = `${sourceName}@@${targetName}`;
+    const prev = linkValueMap.get(key) ?? 0;
+    linkValueMap.set(key, prev + sourceValue);
+  });
+
+  const links: SankeyLink[] = [];
+  const sourceSet = new Set<string>();
+  const targetSet = new Set<string>();
+
+  linkValueMap.forEach((value, key) => {
+    const [source, target] = key.split('@@');
+    if (!source || !target) {
+      return;
+    }
+
+    sourceSet.add(source);
+    targetSet.add(target);
+    links.push({ source, target, value });
+  });
+
+  const nodes: SankeyNode[] = [
+    ...Array.from(sourceSet).map((name) => ({ name, kind: 'source' as const })),
+    ...Array.from(targetSet)
+      .filter((name) => !sourceSet.has(name))
+      .map((name) => ({ name, kind: 'target' as const }))
+  ];
+
+  return {
+    nodes,
+    links,
+    meta: {
+      droppedRows,
+      warnings
+    }
+  };
+}
+
+/**
+ * 用于方向切换：仅交换连线方向，不改动原始聚合结果。
+ */
+export function applyDirection(links: SankeyLink[], direction: DirectionMode): SankeyLink[] {
+  if (direction === 'source-to-target') {
+    return links;
+  }
+
+  return links.map((link) => ({
+    source: link.target,
+    target: link.source,
+    value: link.value
+  }));
+}

src/core/types.ts

@@ -0,0 +1,49 @@
+/**
+ * 统一后的表格结构。
+ * headers 表示首行列名，rows 表示去掉首行后每一行的字符串值。
+ */
+export interface RawTable {
+  headers: string[];
+  rows: string[][];
+}
+
+/**
+ * 用户在界面上配置的列映射规则。
+ */
+export interface MappingConfig {
+  sourceDataColumn: number | null;
+  sourceDescriptionColumns: number[];
+  targetDescriptionColumns: number[];
+  delimiter: string;
+}
+
+/**
+ * 渲染桑基图所需的节点。
+ */
+export interface SankeyNode {
+  name: string;
+  kind: 'source' | 'target';
+}
+
+/**
+ * 渲染桑基图所需的边。
+ */
+export interface SankeyLink {
+  source: string;
+  target: string;
+  value: number;
+}
+
+/**
+ * 聚合后的业务输出，包括告警信息。
+ */
+export interface SankeyBuildResult {
+  nodes: SankeyNode[];
+  links: SankeyLink[];
+  meta: {
+    droppedRows: number;
+    warnings: string[];
+  };
+}
+
+export type DirectionMode = 'source-to-target' | 'target-to-source';