使用shell脚本合并txt文件的行

Question

I invoke a program from shell script and it creates an output file with this format:

aaaaa\
bbbbb\
ccccc\

I would like to change this to:

aaaaabbbbbccccc

In VI editor I can just do ggVGJ and then replace all \\ with "". But I want to get this done via a script.

Answer 1

Here's one way using GNU sed :

sed ':a; N; $!ba; s/\\\n//g; s/\\$//' file

Another way using awk , may give you better performance:

awk '{ sub ("\\\\$", ""); printf "%s", $0 } END { print "" }' file

Results:

aaaaabbbbbccccc

---

Explanation:

The awk solution removes the trailing backslash (via substitution) and printf's each line (without a newline character). END (which is executed at the end of the script) then prints a newline character. This is superior to the sed solution, which creates a label called a and appends the next line of input into the pattern space. $!ba means 'if not at the last line of input, branch to label a '. The first substitution then removes each backslash and newline character from the pattern space. The second substitution removes the last, trailing backslash. This solution should be fast for small files, but probably won't be any faster than the awk for the same file. Although ... it was faster to write.

Answer 2

Here's one way using sed and tr :

sed 's/\\$//' < sample.txt | tr -d '\n'

If you want to add a newline too, you can add an echo at the end:

sed 's/\\$//' < sample.txt | tr -d '\n'; echo

If you want the whole thing to be a one unit, for example to use in a ... && ... || ... ... && ... || ... construct then you can group the two steps like this:

{ sed 's/\\$//' < sample.txt | tr -d '\n'; echo; }

Answer 3

Another way, using pure bash:

$ cat file.txt 
aaaaa\
bbbbb\
ccccc\
$ { cat file.txt ; echo; } | while read line; do echo $line; done
aaaaabbbbbccccc
$

This works because the bash read command actually deals with the \\ continuation automatically (use the -r switch to read to disable this behavior). The echo after the cat is necessary for this example because the last line of your sample text ends in \\ , so the read command doesn't think it has got to the end of a line and doesn't output anything. The echo just inserts an empty line at the end of the stream to clean this up.

Answer 4

I guess this solution is the smallest:

$ cat tmp.txt
aaaaa\
bbbbb\
ccccc\

$ cat tmp.txt | tr -d "\\\r\n"
aaaaabbbbbccccc

Answer 5

试试这条线;

awk -F'\\\\$' '{printf "%s", $1}END{print ""}' file

Answer 6

This is a really ugly hack, but you could use the gcc preprocessor :

 $ cat file.txt 
 aaaaa\
 bbbbb\
 ccccc\
 $ cat file.txt | gcc -xc -E -P -w - | grep .
 aaaaabbbbbccccc
 $ 

Why is this risky? If your input text happened to contain preprocessor directives, then they would get interpreted, resulting in a mess.

Answer 7

One with awk and sed :

sed 's/\\$//g' file | awk '{printf "%s", $1}'

sed command removes the slash at the end of the line. $ denotes the end of the line after a slash. Since slash is considered as a meta character in sed , you need an extra \\ to escape it. piping the output of sed to awk printf prints multiple lines in one. $0 represents the entire line.