nodejs synchronization read large file line by lin

问题:

I have a large file (utf8). I know fs.createReadStream can create stream to read a large file, but not synchronized. So i try to use fs.readSync, but read text is broken like "迈�".

var fs = require('fs');
var util = require('util');
var textPath = __dirname + '/people-daily.txt';   
var fd = fs.openSync(textPath, "r");
var text = fs.readSync(fd, 4, 0, "utf8");
console.log(util.inspect(text, true, null));

回答1:

For large files, readFileSync can be inconvenient, as it loads the whole file in memory. A different synchronous approach is to iteratively call readSync, reading small bits of data at a time, and processing the lines as they come. The following bit of code implements this approach and synchronously processes one line at a time from the file 'test.txt':

var fs = require('fs');
var filename = 'test.txt'

var fd = fs.openSync(filename, 'r');
var bufferSize = 1024;
var buffer = new Buffer(bufferSize);

var leftOver = '';
var read, line, idxStart, idx;
while ((read = fs.readSync(fd, buffer, 0, bufferSize, null)) !== 0) {
  leftOver += buffer.toString('utf8', 0, read);
  idxStart = 0
  while ((idx = leftOver.indexOf("\n", idxStart)) !== -1) {
    line = leftOver.substring(idxStart, idx);
    console.log("one line read: " + line);
    idxStart = idx + 1;
  }
  leftOver = leftOver.substring(idxStart);
}

回答2:

use https://github.com/nacholibre/node-readlines

var lineByLine = require('n-readlines');
var liner = new lineByLine('./textFile.txt');

var line;
var lineNumber = 0;
while (line = liner.next()) {
    console.log('Line ' + lineNumber + ': ' + line.toString('ascii'));
    lineNumber++;
}

console.log('end of line reached');

回答3:

Use readFileSync:

fs.readFileSync(filename, [encoding]) Synchronous version of fs.readFile. Returns the contents of the filename.

If encoding is specified then this function returns a string. Otherwise it returns a buffer.

On a side note, since you are using node, I'd recommend using asynchronous functions.

回答4:

I built a simpler version JB Kohn's answer that uses split() on the buffer. It works on the larger files I tried.

/*
 * Synchronously call fn(text, lineNum) on each line read from file descriptor fd.
 */
function forEachLine (fd, fn) {
    var bufSize = 64 * 1024;
    var buf = new Buffer(bufSize);
    var leftOver = '';
    var lineNum = 0;
    var lines, n;

    while ((n = fs.readSync(fd, buf, 0, bufSize, null)) !== 0) {
        lines = buf.toString('utf8', 0 , n).split('\n');
        lines[0] = leftOver+lines[0];       // add leftover string from previous read
        while (lines.length > 1) {          // process all but the last line
            fn(lines.shift(), lineNum);
            lineNum++;
        }
        leftOver = lines.shift();           // save last line fragment (may be '')
    }
    if (leftOver) {                         // process any remaining line
        fn(leftOver, lineNum);
    }
}

回答5:

two potential problems,

1. 3bytes BOM at the beginning you did not skip
2. first 4bytes cannot be well format to UTF8's chars( utf8 is not fixed length )