[SOLVED] \n newline character in cp command path

mrquartz · 04-13-2015, 12:07 PM

Hi guys,

I am writing a bash script with puppywary 5.3 distro. Basically, the script has to find certain files with extensions and move them to another directory.
Sounds easy, however, if there are files with the same name but different extensions only the one with .docx has to be copied. That's where the fun starts. When I try to do copy command with multiple files I get this:

cp: cannot stat `dir/filename\ndir/filename': no such file or directory

I know that \n is a newline character however I have no idea how does it get in between files when copying and whether its because flaws in code or because I am using distro. It concatenates both paths of files and prevents from copying.

Also, all the files I am copying are empty, using for test.

Thanks guys!

Code:

#!/bin/sh
src=$1
dst=$2
fdoc=$(find "$src" -type f -name "*.doc" 2>/dev/null)
fdocx=$(find "$src" -type f -name "*.docx" 2>/dev/null)
fpdf=$(find "$src" -type f -name "*.pdf" 2>/dev/null)
if [ ! -d "$src" ]; then
   echo "The path: $src does not exist. Please check the path"
   exit=1
fi
[ ! -d "$dst" ]
mkdir -p "$dst"
for i in "$src"; do
if [[ "$(basename "$fdoc" .doc)" == "$(basename "$fdocx" .docx)" ]] 
   cp "$fdocx" "$dst"
   fi
done

Please note: it is just the part of the code to compare two filenames and copy the .docx version.

SoftSprocket · 04-13-2015, 12:17 PM

I would guess there is a bug in your script - of course it is a guess since you haven't posted the script that is causing the error. Should you decide to post the code please use [code][/code] tags.

mina86 · 04-13-2015, 01:07 PM

Code:

fdoc=$(find "$src" -type f -name "*.doc" 2>/dev/null)
fdocx=$(find "$src" -type f -name "*.docx" 2>/dev/null)
fpdf=$(find "$src" -type f -name "*.pdf" 2>/dev/null)

This seems to break once there are multiple .doc, .docx or .pdf files in $src. Your script assumes that if there is any doc or docx file, there is exactly one.

Code:

[ ! -d "$dst" ]

This is pointless (unless you add set -e somewhere at the beginning).

Code:

for i in "$src"; do

This is equivalent to i=$src.

Code:

if [[ "$(basename "$fdoc" .doc)" == "$(basename "$fdocx" .docx)" ]]

Youre claiming to use /bin/sh so do so:

Code:

if [ "$(basename "$fdoc" .doc)" = "$(basename "$fdocx" .docx)" ]

mrquartz · 04-13-2015, 01:16 PM

Quote:

Originally Posted by mina86

Code:

fdoc=$(find "$src" -type f -name "*.doc" 2>/dev/null)
fdocx=$(find "$src" -type f -name "*.docx" 2>/dev/null)
fpdf=$(find "$src" -type f -name "*.pdf" 2>/dev/null)

This seems to break once there are multiple .doc, .docx or .pdf files in $src. Your script assumes that if there is any doc or docx file, there is exactly one.

Hey mina86 and thank you for your reply.
Any ideas how to fix my code ? I am relatively new to bash scripting.

However, when I use find command it gives me all files with .doc .docx or .pdf extensions.

mina86 · 04-13-2015, 01:33 PM

I usually just implement find in shell if I need to go through all files (with given extension) in a given directory and all its subdirectories. If you care about bash only, you can use extended globs like **/*.docx.

mrquartz · 04-13-2015, 01:37 PM

Thanks for the idea, however, I believe problem is not in this this.

mina86 · 04-13-2015, 02:08 PM

Im pretty sure the way you use find is the problem.

mrquartz · 04-13-2015, 02:34 PM

Could you then write the find, how you think it should be ?

michaelk · 04-13-2015, 03:24 PM

Not sure I understand the problem. If the goal is to find all *.docx files and copy them to another directory then I do not understand the significance if there are also *.doc or *.pdf files of the same name or the necessity to compare file names.

Homework assignment?

mrquartz · 04-13-2015, 03:50 PM

Quote:

Originally Posted by michaelk

Not sure I understand the problem. If the goal is to find all *.docx files and copy them to another directory then I do not understand the significance if there are also *.doc or *.pdf files of the same name or the necessity to compare file names.

Homework assignment?

Hey MichaelK and thanks for joining us.

The problem is to find all .doc, .docx and .pdf files and copy them to a directory. However, if names of files with .doc and .docx are the same then only .docx should be copied.

Yes it is! haha, how did you figure out that

Hope this makes it clearer.

grail · 04-13-2015, 07:58 PM

Well the first thing that stood out to me was this:

Code:

if [ ! -d "$src" ]; then
   echo "The path: $src does not exist. Please check the path"
   exit=1
fi

You call this 'after' having already attempting to use $src in your find commands ... bit pointless after the fact.

Then as the question is homework, my suggestion would be:

1. Use find with -exec option to copy pdf and docx files to location

2. Pump find of doc files into a loop (while) and test the file name does not exist in the destination and then copy if not there

Note: Your current issue is with find as the default output from the find command is to place each found item on a separate line using newline. You could look into the -print option of find

michaelk · 04-13-2015, 08:05 PM

hint:
http://www.ibm.com/developerworks/ai...unix-find.html

NevemTeve · 04-13-2015, 11:21 PM

example

Code:

find . -iname '*.doc' -print | while read -r FILE; do
    process $FILE
done

mina86 · 04-14-2015, 06:55 AM

Code:

find . -iname '*.doc' -print | while read -r FILE; do

Which of course breaks if there is a file with newline in its name, which is the reason I suggest implementing find in bash.

Quote:

Originally Posted by mrquartz

The problem is to find all .doc, .docx and .pdf files and copy them to a directory. However, if names of files with .doc and .docx are the same then only .docx should be copied.

What if there is file with the same name in two different subdirectories of $src?

NevemTeve · 04-14-2015, 07:50 AM

> Which of course breaks if there is a file with newline in its name, which is the reason I suggest implementing find in bash.

Yes, those who have NL's in the file-names are in trouble...

Code:

find . -name '*.docx' -exec somecript {} \;