Bash中没有空格的字符串连接

Question

I'm trying to write a simple bash script to count the number of files in a directory, and then add a new file with name file<#files> to the end of the directory. My current attempt is:

name="out"
num=$(ls -l|wc -l)
echo foo > "${name}${num}"

However, this gives me a bunch of spaces, resulting in the filename out 12 . Why do the spaces appear, and how do I concatenate these strings without creating spaces?

Answer 1

On OS X (and BSD-like systems in general) , wc -l left-space-pads the number to 8 characters , eg, _______7 ( _ representing a space here for technical reasons).

(If you use wc 's output unquoted with echo - eg, echo $(wc -l <<<'dummy') , you will not see the padding, because the shell will "eat" the leading spaces; if you double-quote the command substitution, you'll see them: echo "$(wc -l <<<'dummy')"

That said, you're better off using neither ls nor wc - for both reasons of robustness and performance ; use globbing to capture all filenames in an array and use that array's element count instead:

name='out'
files=( * ) # collect all filenames in array

echo foo > "${name}${#files[@]}" # use array-element count

^{Note: By default, * will not include hidden items, just like the OP's ls command (because it doesn't include -A or -a ).
Use shopt -s dotglob to include hidden items too.
Also, if there happen to be no matching items at all, Bash will return the pattern as-is , ie, * ; to have Bash return the empty string instead, use shopt -s nullglob .}

---

Generally , if you do find yourself needing to trim leading and trailing whitespace from command output and/or a Bash variable :

• To trim a value stored in a Bash variable , use read :

    # Single-line value: val=' ab ' read -r val <<<"$val" # $val now contains 'ab' # Multi-line value: # Note: Trims leading and trailing whitespace including newlines, but # preserves *any* interior whitespace, including empty and all-whitespace lines. val=$'\\n \\t one\\n \\ntwo\\n \\n' read -r -d '' val <<<"$val" # $val now contains $'one\\n \\ntwo' 

• To trim command output as part of a pipeline or lines from stdin / a file :

  • If the value to trim has either no interior whitespace or you want to ensure / don't mind that interior whitespace is normalized to a single space each , pipe to xargs (this works, because xargs , after parsing the input into words, passes them to the echo utility by default - without shell involvement); the result is always a single output line ; caveat : xargs removes embedded quotes and \\ instances, unless you \\ -escape them:

     val=$(ls | wc -l | xargs) # $val now contains trimmed count 

  • If preserving line-interior whitespace as-is matters , use sed ; note that - unlike read -rd '' above - this is a LINE-based solution, so the number of input lines will be preserved , including leading and trailing empty or all-whitespace lines, with all-whitespace lines trimmed to empty ones (assumes GNU Sed or BSD Sed):

     echo $'\\nfoo \\n \\n bar \\n' | sed -E 's/^[[:blank:]]+|[[:blank:]]+$//g' # -> $'\\nfoo\\n\\nbar' 

---

• Simplified scenario: Remove ALL whitespace from a value, using tr :

  • single-line / per-line :

     tr -d '[:blank:]' <<<$' abc \\n foo ' # -> $'abc\\nfoo\\n' 

  • globally, including newlines :

     tr -d '[:space:]' <<<$' abc \\n foo ' # -> 'abcfoo' 

Answer 2

http://mywiki.wooledge.org/BashFAQ/004 describes how to concisely count the number of items in a directory. For example, the following code might do what you want:

shopt -s nullglob dotglob
items=(*)
touch "file${#items[@]}"
echo "${#items[@]} items found, created new file: file${#items}"

Please note, however, that this count includes subdirectories and symlinks, not just files. In your description you asked for a count of files, though in your code example you seem to also want subdirectories and such. If you only want files, then the above example will have to be modified, perhaps with something like 'find . -type f'.

Answer 3

try this:

#!/bin/sh
name="out"
num=$(ls -l | wc -l | sed -e 's/ //g')
echo $num

which should strip out the beginning whitespace.