带重定向的 tee bash 命令

Question

I have the following file:

file1.txt

geek
for
geeks

I am using the tee command to perform two operations on the output. My question is about the the redirection character after the first tee . I want to get the first column of file1.txt and write it to file2.txt. When I run the following command, I don't receive an error but it does not give me the first column:

wc -l file1.txt |tee awk '{print $1}' - > file2.txt | sed 's/4/6/g' > file3.txt

However, the following command works as expected. What does the > is doing here?

wc -l file1.txt |tee >(awk '{print $1}' - > file2.txt) | sed 's/4/6/g' > file3.txt

Answer 1

tee awk '{print $1}' - > file2.txt

does:

• execute tee with 3 arguments awk and '{print $1}' and - .
• tee will create a file named awk , another file named '{print $1}' and yet another file named - .
• Then the output of tee will be redirected to file2.txt
• tee will duplicate input to those 3 files and will output to file2.txt
• Consequently | sed | sed will receive no input, because the output of tee is redirected to the file and the subshell outputs nothing.

---

 tee >(awk '{print $1}' - > file2.txt)

does:

• >(...)
  • Run awk with two arguments '{print $1}' and -
  • ------ '{print $1}' is interpreted as a script
  • ------ - is interpreted as stdin (and could be omitted)
  • ------ then the output of awk of redirected to file2.txt
  • Then bash creates a fifo or a /dev/fd/something file
  • Then the output of that file is connected to stdin of awk process
  • And the >(awk ...) is substituted for the filename of the file, most probably for /dev/fd/something
• tee >(...)
  • executes tee with one argument, like tee /dev/fd/something
  • The /dev/fd/something is connected to awk process on the other side
  • So tee writes to /dev/fd/something and awk reads the data from stdin on the other side
  • the output of tee is redirected to | sed | sed

---

What does the > is doing here?

The first occurrence is used to introduce a process substitution. The second occurrence is used to redirect output of awk command to a file named file2.txt . The third occurrence is used to redirect the output of sed command to file named file3.txt .

Answer 2

Here, Process substitution is used to capture output that would normally go to a file

The Bash syntax for writing to a process is >(command)