refactor(ai-chat): 移除 excel_analyze/office_analyze 工具与 OfficeKit 依赖

- 删除 excel_analyze/office_analyze 两个 A2UI 工具(schema、分发、执行器、测试)
- 附件分析改为依赖上传时自动提取的文本与 preflight_import 报告 errorSamples
- 移除 @officecli/officecli 依赖及 OfficeCliService(PPT 上传不再自动提取文本)
- 保留 AiExcelReaderService(附件文本提取与导入预检仍使用)
This commit is contained in:
2026-08-05 22:26:05 +08:00
parent d1c933f032
commit 6249fefc64
15 changed files with 2 additions and 720 deletions

View File

@@ -8,11 +8,9 @@ import { createReadStream } from 'node:fs';
import { mkdir, readFile, unlink, writeFile } from 'node:fs/promises';
import { basename, isAbsolute, join, relative, resolve } from 'node:path';
import { randomUUID } from 'node:crypto';
import { tmpdir } from 'node:os';
import { PDFParse } from 'pdf-parse';
import { In, Repository } from 'typeorm';
import { AiExcelReaderService } from './ai-excel-reader.service';
import { OfficeCliService } from './office-cli.service';
import { AiAttachment } from './entities';
const MAX_FILE_BYTES = 10 * 1024 * 1024;
@@ -42,7 +40,6 @@ export class AiAttachmentService {
@InjectRepository(AiAttachment)
private readonly attachments: Repository<AiAttachment>,
private readonly excelReader: AiExcelReaderService,
private readonly officeCli?: OfficeCliService,
) {}
async upload(userId: number, file: Express.Multer.File): Promise<AiAttachment> {
@@ -213,68 +210,12 @@ export class AiAttachmentService {
const result = await mammoth.extractRawText({ buffer });
return this.normalizeExtractedText(result.value);
}
if (mimeType.includes('presentationml')) {
if (!this.officeCli) return null;
const text = await this.extractWithOfficeCli(buffer, mimeType);
return this.normalizeExtractedText(text);
}
if (mimeType.includes('spreadsheetml')) {
return this.normalizeExtractedText(await this.excelReader.extractText(buffer));
}
return null;
}
private async extractWithOfficeCli(buffer: Buffer, mimeType: string): Promise<string> {
if (!this.officeCli) return '';
const extension = this.extensionForMime(mimeType);
const tempPath = join(tmpdir(), `${randomUUID()}.${extension}`);
try {
await writeFile(tempPath, buffer, { flag: 'wx' });
const result = await this.officeCli.view(tempPath, 'text');
if (!result.success || !result.data || typeof result.data !== 'object') return '';
const data = result.data as { sheets?: Array<{ name: string; rows: unknown[] }>; elements?: Array<{ text?: string }> };
if (Array.isArray(data.sheets)) {
return data.sheets
.map((sheet) => {
const lines: string[] = [];
for (const row of sheet.rows ?? []) {
if (!row || typeof row !== 'object' || !('cells' in row)) continue;
const cells = (row as { cells: Record<string, unknown> }).cells;
const placed = new Map<number, string>();
let maxColumn = -1;
for (const [key, value] of Object.entries(cells)) {
const columnIndex = this.officeColumnIndex(key.replace(/\d+/g, ''));
placed.set(columnIndex, String(value ?? ''));
maxColumn = Math.max(maxColumn, columnIndex);
}
if (maxColumn < 0) continue;
const line = Array.from({ length: maxColumn + 1 }, (_, index) => placed.get(index) ?? '').join('\t');
if (line.trim()) lines.push(line);
}
return `# ${sheet.name}\n${lines.join('\n')}`;
})
.join('\n');
}
if (Array.isArray(data.elements)) {
return data.elements
.map((element) => element.text ?? '')
.filter((line) => line.trim() !== '')
.join('\n');
}
return '';
} finally {
await unlink(tempPath).catch(() => undefined);
}
}
private officeColumnIndex(letters: string): number {
let index = 0;
for (const char of letters.toUpperCase()) {
index = index * 26 + (char.charCodeAt(0) - 64);
}
return index - 1;
}
/**
* Read the stored file content of an already-owned attachment so the AI
* chat agent can page through large workbooks on demand.