等待在 Bash 中创建文件

Waiting for a file to be created in Bash

提问by J.Clark

I need to create a bash script to wait for a file to be created. The script will use sleep command inside a while loop to periodically check on a file every 10 seconds. Print out a message while waiting. Display the content of the file once the file is created. Below is what I have tried to implement and it obviously does not work. At this point, I'm not entirely sure how to proceed.

我需要创建一个 bash 脚本来等待创建文件。该脚本将在 while 循环中使用 sleep 命令每隔 10 秒定期检查文件。等待时打印一条消息。创建文件后显示文件的内容。以下是我尝试实施的内容，但显然行不通。在这一点上，我不完全确定如何进行。

#!/bin/bash
let file=

while '( -f !  /tmp/)'
do
       sleep 10
       echo "still waiting"
done

echo "Content of the file :"

回答by Charles Duffy

The problem here is with the test, not the sleep (as the original question hypothesized). The smallest possible fix might look as follows:

这里的问题在于测试，而不是睡眠（正如最初的问题假设的那样）。最小的修复可能如下所示：

while ! test -f "/tmp/"; do
  sleep 10
  echo "Still waiting"
done

---

---

Keep in mind the syntax for a whileloop:

请记住while循环的语法：

while: while COMMANDS; do COMMANDS; done
    Expand and execute COMMANDS as long as the final command in the
    `while' COMMANDS has an exit status of zero.

while: while COMMANDS; do COMMANDS; done
    Expand and execute COMMANDS as long as the final command in the
    `while' COMMANDS has an exit status of zero.

That is to say, the first argument given to while, expanding the loop, is a command; it needs to follow the same syntax rules as any other shell command.

也就是说，给 的第一个参数while，展开循环，是一个命令；它需要遵循与任何其他 shell 命令相同的语法规则。

-fis valid as an argument to test-- a command which is also accessible under the name [, requiring a ]as the last argument when used in that name -- but it's not valid as a command in and of itself -- and when passed as part of a string, it's not even a shell word that couldbe parsed as an individual command name or argument.

-f作为参数有效test- 也可以在 name 下访问的命令，在该名称中使用[时需要 a]作为最后一个参数 - 但它本身作为命令无效 - 当作为的一部分传递时一个字符串，它甚至不是一个shell词，可以被解析为一个单独的命令名称或参数。

When you run '( -f ! /tmp/$1)'as a command, inside quotes, the shell is looking for an actual command with exactly that name (including spaces). You probably don't have a file named '/usr/bin/( -f ! /tmp/$1)'in your PATH or any other command by that name found, so it'll always fail -- exiting the whileloop immediately.

当您'( -f ! /tmp/$1)'作为命令运行时，在引号内，shell 正在寻找具有该名称（包括空格）的实际命令。您可能没有'/usr/bin/( -f ! /tmp/$1)'在您的 PATH 中命名的文件或任何其他以该名称命名的命令，因此它总是会失败——while立即退出循环。

---

---

By the way -- if you're willing to make your code OS-specific, there are approaches other than using sleepto wait for a file to exist. Consider, for instance, inotifywait, from the inotify-toolspackage:

顺便说一句——如果你愿意让你的代码特定于操作系统，除了sleep用来等待文件存在之外，还有其他方法。例如inotifywait，从inotify-tools包中考虑：

while ! test -f "/tmp/"; do
  echo "waiting for a change to the contents of /tmp" >&2
  inotifywait --timeout 10 --event create /tmp >/dev/null || {
    (( $? == 2 )) && continue  ## inotify exit status 2 means timeout expired
    echo "unable to sleep with inotifywait; doing unconditional 10-second loop" >&2
    sleep 10
  }
done

The benefit of an inotify-based interface is that it returns immediately upon a filesystem change, and doesn't incur polling overhead (which can be particularly significant if it prevents a system from sleeping).

基于 inotify 的接口的好处是它在文件系统更改时立即返回，并且不会产生轮询开销（如果它阻止系统休眠，这可能特别重要）。

---

---

By the way, some practice notes:

顺便说一下，一些练习笔记：

• Quoting expansions in filenames (ie. "/tmp/$1") prevents names with spaces or wildcards from being expanded into multiple distinct arguments.
• Using >&2on echocommands meant to log for human consumption keeps stderr available for programmatic consumption
• letis used for math, not general-purpose assignments. If you want to use "$file", nothing wrong with that -- but the assignment should just be file=$1, with no preceding let.

• 在文件名中引用扩展（即"/tmp/$1"）可防止带有空格或通配符的名称被扩展为多个不同的参数。
• 使用>&2上echo意味着日志供人食用的命令保持标准错误可以编程消费
• let用于数学，而不是通用任务。如果你想使用"$file"，那没有错——但赋值应该是file=$1，没有前面的let。