Filesystem 函数
在线手册：中文 英文
PHP手册

fscanf

(PHP 4 >= 4.0.1, PHP 5)

fscanf — 从文件中格式化输入

说明

mixed fscanf ( resource $handle , string $format [, mixed &$... ] )

fscanf() 函数和 sscanf() 相似，但是它从与 handle 关联的文件中接受输入并根据指定的 format（定义于 sprintf() 的文档中）来解释输入。如果只给此函数传递了两个参数，解析后的值会被作为数组返回。否则，如果提供了可选参数，此函数将返回被赋值的数目。可选参数必须用引用传递。

格式字符串中的任何空白会与输入流中的任何空白匹配。这意味着甚至格式字符串中的制表符 \t 也会与输入流中的一个空格字符匹配。

Example #1 fscanf() 例子

<?php
$handle = fopen("users.txt","r");
while ($userinfo = fscanf($handle, "%s\t%s\t%s\n")) {
    list ($name, $profession, $countrycode) = $userinfo;
    //... do something with the values
}
fclose($handle);
?>

Example #2 users.txt 的内容

javier  argonaut        pe
hiroshi sculptor        jp
robert  slacker us
luigi   florist it

Note: 在 PHP 4.3.0 之前，从文件中读入的最大字符数是 512（或者第一个 \n，看先碰到哪种情况）。从 PHP 4.3.0 起可以读取任意长的行。

参见 fread()，fgets()，fgetss()，sscanf()，printf() 和 sprintf()。

Filesystem 函数
在线手册：中文 英文
PHP手册
PHP手册 - N: 从文件中格式化输入

用户评论:

dave at dave dot st (01-Jan-2009 07:58)

The fgetd() function suggested by worldwideroach on 14-Jul-2005 04:33 does not handle the last buffer-load of data correctly. At least not for my requirement. It is possible for EOF to have been reached but for there still to be characters in $sRecord.

For the, er... record I got better results with this version:

<?php
function fgetd(&$rFile, $sDelim, $iBuffer=1024) {

  $sRecord = '';

  while(!feof($rFile)) {
    $iPos = strpos($sRecord, $sDelim);
    if ($iPos === FALSE) {
        $sRecord .= fread($rFile, $iBuffer);
    } else {
        fseek($rFile, 0-strlen($sRecord)+$iPos+strlen($sDelim), SEEK_CUR);
        return substr($sRecord, 0, $iPos);
    }
  }

  // Last read got some more data before hitting EOF?
  if ($sRecord != '') {
    if (($iPos = strpos($sRecord, $sDelim)) !== FALSE) {
      fseek($rFile, 0-strlen($sRecord)+$iPos+strlen($sDelim), SEEK_CUR);
      return substr($sRecord, 0, $iPos);
    }
    else {
      return $sRecord;
    }
  }
  else {
    return FALSE;
  }
}
?>

arentzen at religion dot dk (30-Sep-2007 10:56)

If you want fscanf()to scan one variable in a large number of lines,  e.g an Ipadress in a line with more variables, then use fscanf with explode()
<?
$filename = "somefile.txt";
$fp = fopen($filename, "r") or die ("Error opening file! \n");
$u = explode(" ",$line); // $u is the variable eg. an IPadress
while ($line = fscanf($fp,"%s",$u)) {
if(preg_match("/^$u/",$_SERVER['REMOTE_ADDR'])) {$badipadresss++;} // do something and continue scan
}
?>
Besides, fscanf()is much faster than fgets()

Bertrand dot Lecun at prism dot uvsq dot Fr (30-May-2007 08:48)

It would be great to precise in the fscanf documentation
that one call to the function, reads a complete line.
and not just the number of values defined in the format.

If a text file contains 2 lines each containing 4 integer values,
reading the file with 8 fscanf($fd,"%d",$v) doesnt run !
You have to make 2
fscanf($fd,"%d %d %d %d",$v1,$v2,$v3,$v4);

Then 1 fscanf per line.

loco.xxx at gmail dot com (24-Jul-2006 08:46)

to include all type of visible chars you should try:

<?php fscanf($file_handler,"%[ -~]"); ?>

worldwideroach at hotmail dot com (14-Jul-2005 05:33)

Yet another function to read a file and return a record/string by a delimiter.  It is very much like fgets() with the delimiter being an additional parameter.  Works great across multiple lines.

function fgetd(&$rFile, $sDelim, $iBuffer=1024) {
    $sRecord = '';
    while(!feof($rFile)) {
        $iPos = strpos($sRecord, $sDelim);
        if ($iPos === false) {
            $sRecord .= fread($rFile, $iBuffer);
        } else {
            fseek($rFile, 0-strlen($sRecord)+$iPos+strlen($sDelim), SEEK_CUR);
            return substr($sRecord, 0, $iPos);
        }
    }
    return false;
}

rudigreen at gmail dot com (01-Jul-2005 11:43)

I have a function for reading delimited files, it works for multiple lines too (i think...)

<?
//$fh - is the file pointer
//$delim - is the seperator
//$callback - self explanatory
//$len - optional
function file_read_delim($fh,$delim,$callback,$len=1024)
{
    $rec = '';
    while(!feof($fh))
    {
        $buf = fread($fh,$len);
        if(strpos($buf,$delim) === false)
        {
            $rec .= $buf;
        }
        else
        {
            $strs = explode($delim,$buf);
            foreach ($strs as $ele)
            {
                $rec .= $ele;
                call_user_func($callback,$rec);
                $rec = '';
            }
        }
    }
}

//Here is an example how to use the function

$fh = fopen($filename,'r');
    if(!$fh)
    {
                 die 'Could not open file for reading';
    }
        //call the function
    file_read_delim($fh,'-','cb');
    fclose($fh);

function cb($rec)
{
echo "$rec \n";
}
?>

me at hesterc dot fsnet dot co dot uk (25-May-2004 03:03)

I have a simpler method I use to parse delimited text. Using the data posted by gozer at fanhunter dot com, here is my script. Maybe it is faster?

<?php

$fp = fopen ("sections.dat","r");

if (!$fp) {echo "<p>Unable to open remote file.</p>"; exit;}

while (!feof($fp)):
 $line = fgets($fp, 2048);
 $out = array($line);
 list ($id, $name, $description, $language, $directory, $id_uplevel, $order, $hassubsection) = split ("\|", $out[0]);
 echo "$id-$name-$description-$language-$directory-
$id_uplevel-$order-$hassubsection<br />\n";
 $fp++;
endwhile;

fclose($fp);
?>

Notes:

Avoid the php extension on a data file - it will cause PHP to parse the file, but there is no PHP in it.

The "2048" value on line 2 of the loop is set for long lines. 1024 works fine, but I had to increase it with a large database I use a similar script to read.

You don't need to open and close the speech marks (as in gozer at fanhunter dot com's example) in the echo line, just use the variables inbetween the dashes.

(Remove the line break halfway through the echo line - it is just there for this forum.)

matt at mattsinclair dot com (21-Jan-2004 08:36)

A better way to use fscanf() would be this:

<?php
$handle = fopen("users.txt", "r");
while (!feof($handle)) {
   $userinfo = fscanf($handle, "%s\t%s\t%s\n");
   if ($userinfo) {
     list ($name, $profession, $countrycode) = $userinfo;
     //... do something with the values
   }
   $userinfo=NULL;
}
fclose($handle);
?>

as you can see, instead of waiting for fscanf() to fail to return a value... it waits for the the pointer to get to the end of the file... this way, if for some reason one of your lines does not match your expression, it will not kill the loop.  it will simply go on to the next line.

robert at NOSPAM dot NOSPAM (25-Oct-2002 12:08)

actually, instead of trying to think of every character that might be in your file, excluding the delimiter would be much easier.

for example, if your delimiter was a comma use:

%[^,]

instead of:

%[a-zA-Z0-9.| ... ]

Just make sure to use %[^,\n] on your last entry so you don't include the newline.

ruiner911 at yahoo dot com (15-Aug-2002 09:01)

Clear the variables before you scan them in.  As a programmer this should have been very apparent.  Goof.

eugene at pro-access dot com (16-Mar-2002 08:39)

If you want to read text files in csv format or the like(no matter what character the fields are separated with), you should use fgetcsv() instead. When a text for a field is blank, fscanf() may skip it and fill it with the next text, whereas fgetcsv() correctly regards it as a blank field.

gozer at fanhunter dot com (07-Mar-2002 02:53)

Hi,
A few days ago we got multiple mySQL crashes due to a hardware failure and other processes running.
While we thought it could be the mySQL daemon overloaded, we started looking for alternate ways to get our little databases working so we started using fscanf to parse files.

We ran into multiple problems due to the whitespace and other characters that were in our database. Finally, we made it to work using sets as james@zephyr-works.com remarked.

Our final function is:

  function get_sections($include_dir){
    $filename = $include_dir . "sections.dat.php";
    $datafile = fopen ($filename ,"r");
    while ($sectioninfo = fscanf ($datafile, "%[0-9]|%[a-zA-Z0-9@&;:,. /!?-]|%[a-zA-Z0-9@&;:,. /!?-]|%[a-zA-Z]|%[a-zA-Z0-9@/?&;.+=-]|%[0-9]|%[0-9]|%[0-9]\n")) {
        list($id, $name, $description, $language, $directory, $id_uplevel, $order, $hassubsection) = $sectioninfo;

        // Show output
        echo $id . "-" . $name. "-" . $description . "-" . $language . "-" . $directory . "-" . $id_uplevel . "-" . $order . "-" . $hassubsection . "<br>\n";
    }
    fclose($datafile);
  }

The contents of sections.dat.php (for example):

1|home|P&aacute;gina principal de Fanhunter.|castellano|==|0|0|0
2|fanhunter|Secci&oacute;n principal dedicada al universo Fanhunter.|castellano|fanhunter/|1|0|0
3|outfan|Secci&oacute;n principal dedicada al universo Outfan.|castellano|outfan/|1|0|0
4|fanpiro|Secci&oacute;n principal dedicada al universo Fanpiro.|castellano|fanpiro/|1|0|0
5|tienda|La tienda de Fanhunter.|castellano|tienda/|1|0|0
6|the zone|Secci&oacute;n principal Miscel&aacute;nea.|castellano|thezone/|1|0|0
7|flfcn|Secci&oacute;n principal dedicada a Fan Letal/Fan con Nata.|castellano|fanletal/|1|0|0
8|foro|Nuestro foro de discusi&oacute;n.|castellano|foro/|1|0|0
9|chat|Secci&oacute;n para chatear.|castellano|chat/|1|0|0
10|links|Secci&oacute;n recopilatoria de enlaces de inter&eacute;s a otras p&aacute;ginas.|castellano|links/|1|0|0

Note: The '==' in directory means no directory needed to be specified.
Pay attention to linebreaks, as this forum puts some of them into the code I pasted.

Good luck guys.

james at zephyr-works dot com (08-Jul-2001 08:29)

fscanf works a little retardedly I've found. Instead of using just a plain %s you probably will need to use sets instead. Because it works so screwy compared to C/C++, fscanf does not have the ability to scan ahead in a string and pattern match correctly, so a seemingly perfect function call like:

fscanf($fh, "%s::%s");

With a file like:

user::password

Will not work. When fscanf looks for a string, it will look and stop at nothing except for a whitespace so :: and everything except whitespace is considered part of that string, however you can make it a little smarter by:

fscanf($fh, "%[a-zA-Z0-9,. ]::%[a-zA-Z0-9,. ]" $var1, $var2);

Which tells it that it can only accept a through z A through Z 0 through 9 a comma a period and a whitespace as input to the string, everything else cause it to stop taking in as input and continue parsing the line. This is very useful if you want to get a sentence into the string and you're not sure of exactly how many words to add, etc.

yasuo_ohgaki at hotmail dot com (13-Mar-2001 07:59)

For C/C++ programmers.

fscanf() does not work like C/C++, because PHP's fscanf() move file pointer the next line implicitly.