Node.js/TypeScript读取大文本文件时坐标解析出现Null问题求助

你的问题确实是流的chunk处理导致的：Node.js文件流是按字节块读取的，一个chunk的结尾可能刚好截断某一行的内容。比如某一行401440.00 5650223.00 344.87可能被拆成两个部分：第一个chunk结尾是401440.0，第二个chunk开头是0 5650223.00 344.87。split('\n')之后就会得到不完整的行，解析时+line[0]会变成NaN（你看到的null应该是NaN的误判，解析失败时得到的是NaN而非null）。

• 临时打印chunk的首尾内容，确认是否存在行截断：

  stream.on('data', chunk => {
    const chunkStr = chunk.toString();
    console.log('Chunk start:', chunkStr.slice(0, 20));
    console.log('Chunk end:', chunkStr.slice(-20));
    // ... 原有代码
  })
  

• 在解析前检查行的有效性，打印异常行：

  .map((line: string[]) => {
    if (line.length < 3 || isNaN(+line[0]) || isNaN(+line[1]) || isNaN(+line[2])) {
      console.log('Invalid line content:', line);
      return;
    }
    // ... 原有解析逻辑
  })
  

需要维护一个剩余行缓冲区，把上一个chunk未完成的行和当前chunk内容拼接后再处理完整行：

new Promise((resolve, reject) => {
  const stream = createReadStream(`${folder}/${file}`);
  let leftover = ''; // 存储上一个chunk未完成的行

  stream.on('error', function(error) {
    const msg = `error: ${error.message}`;
    console.log(msg);
    reject(msg);
  });

  stream.on('data', chunk => {
    const chunkStr = leftover + chunk.toString();
    const lines = chunkStr.split('\n');
    // 把最后一行（可能不完整）存回缓冲区
    leftover = lines.pop() || '';

    lines.forEach(line => {
      if (!line.trim()) return; // 跳过空行
      const parts = line.split(/[,.\s;]+/);
      if (parts.length < 3) return; // 跳过无效行

      const x = +parts[0];
      const y = +parts[1];
      const z = +parts[2];
      
      if (isNaN(x) || isNaN(y) || isNaN(z)) {
        console.log('Skipping invalid point:', line);
        return;
      }

      const point = [x, y, z];
      if (isInsideBoundingBox(point, bbox, bboxSize)) {
        coordinatesList.push({ x, y, z });
      }
    });
  });

  stream.on('close', () => {
    // 处理最后剩余的完整行
    if (leftover.trim()) {
      const parts = leftover.split(/[,.\s;]+/);
      if (parts.length === 3) {
        const x = +parts[0];
        const y = +parts[1];
        const z = +parts[2];
        if (!isNaN(x) && !isNaN(y) && !isNaN(z)) {
          const point = [x, y, z];
          if (isInsideBoundingBox(point, bbox, bboxSize)) {
            coordinatesList.push({ x, y, z });
          }
        }
      }
    }
    resolve('');
  });
});

用Node.js内置的readline模块，它已经封装了行处理逻辑，自动解决chunk截断问题：

import * as readline from 'readline';
import { createReadStream } from 'fs';

new Promise((resolve, reject) => {
  const rl = readline.createInterface({
    input: createReadStream(`${folder}/${file}`),
    crlfDelay: Infinity // 兼容所有换行符格式
  });

  rl.on('line', (line) => {
    if (!line.trim()) return;
    const parts = line.split(/[,.\s;]+/);
    if (parts.length < 3) return;

    const x = +parts[0];
    const y = +parts[1];
    const z = +parts[2];
    if (isNaN(x) || isNaN(y) || isNaN(z)) {
      console.log('Skipping invalid point:', line);
      return;
    }

    const point = [x, y, z];
    if (isInsideBoundingBox(point, bbox, bboxSize)) {
      coordinatesList.push({ x, y, z });
    }
  });

  rl.on('close', () => resolve(''));
  rl.on('error', (error) => {
    const msg = `error: ${error.message}`;
    console.log(msg);
    reject(msg);
  });
});

内容的提问来源于stack exchange，提问作者pcace