Sed search and replace

Dev Shed Forums Sponsor:

September 21st, 2006, 03:47 PM

jabs

Registered User

Join Date: Feb 2004

Posts: 19

Time spent in forums: 3 h 51 m 26 sec
Reputation Power: 0

Sed search and replace

Hi all,

I want to write a sed or awk routine that will find the instance where a line feed and double quote are together in a line and replace them with just the " double quote. I can replace all line feeds, for one gigantic line of data, but that's no good either...I'm not having much success.

Here is my brilliant, errrr, not so successful code to date. http://forums.devshed.com/newthread.php?do=newthread&f=35#
eh?

sed -e 's/"\012/"/g' File1 > File2

TIA for any help.

joe.

September 27th, 2006, 07:57 AM

stanleypane

Contributing User

Join Date: Oct 2003

Location: Baltimore, MD

Posts: 208

Time spent in forums: 4 Days 7 h 4 m 19 sec
Reputation Power: 21

Quote:

Originally Posted by jabs

You'll want to use awk for this purpose. Sed doesn't doesn't handle newlines. Try this:

Code:

awk '/\"$/{printf "%s",$0; next}{print}' file1 > file2

September 28th, 2006, 12:29 PM

jabs

Registered User

Join Date: Feb 2004

Posts: 19

Time spent in forums: 3 h 51 m 26 sec
Reputation Power: 0

Quote:

Originally Posted by stanleypane

You'll want to use awk for this purpose. Sed doesn't doesn't handle newlines. Try this:

Code:

awk '/\"$/{printf "%s",$0; next}{print}' file1 > file2

Thanks for the help. Unfortuntely, it made the file one huge line.

Here is an example of the data I'm using:

1234|320|1|"Sample of data"
1234|321|2|"for DevShed"
these are good.

Here are the bad lines
9876|1000|1|"This sample
"
9876|1001|2|"show the bad rec
"
56574|1015|1|"Another bad one
"

For the good data, obviously, I'd like it to remain the same. For the bad, I'd like to move the " to the end of the preceding line (or add a " to the end of line 1 and delete line 2).

Thanks again for looking at this!

Joe.

September 28th, 2006, 02:10 PM

stanleypane

Contributing User

Join Date: Oct 2003

Location: Baltimore, MD

Posts: 208

Time spent in forums: 4 Days 7 h 4 m 19 sec
Reputation Power: 21

Ahhhh... I understand your problem a little better now. Sorry for the confusion.

If you know that all offending lines always begin with a doube quote, then you can simply remove those lines via grep. Then, any lines that don't end with a double quote can have it added.

There's probably a billion ways to do this, but here goes two:

Code:

Method 1 - grep & sed

   grep "^[^\"]" file1 | sed -e "s/\([^\"]\)$/\1\"/g" > file2

Method 2 - grep & awk

   grep "^[^\"]" file1 | awk '/[^\"]$/{printf "%s\"\n",$0; next}{print}' > file2

Hope this helps!

September 28th, 2006, 02:13 PM

Ehlanna

Not a clue what to put ...

Join Date: Jan 2006

Location: in front of this keyboard

Posts: 815

Time spent in forums: 2 Weeks 2 Days 5 h 43 m 48 sec
Reputation Power: 243

Change it to %s\n in the printf to add a newline in.

__________________
According to Sod's Law, buttered toast lands butter side down, when dropped.
Per nature, cats always land on their feet.
So, what happens when you strap buttered toast to the back of a cat and throw it out a window?.

September 30th, 2006, 08:52 AM

jabs

Registered User

Join Date: Feb 2004

Posts: 19

Time spent in forums: 3 h 51 m 26 sec
Reputation Power: 0

Quote:

Originally Posted by stanleypane

Code:

Method 1 - grep & sed

   grep "^[^\"]" file1 | sed -e "s/\([^\"]\)$/\1\"/g" > file2

Method 2 - grep & awk

   grep "^[^\"]" file1 | awk '/[^\"]$/{printf "%s\"\n",$0; next}{print}' > file2

Hope this helps!

Yes - The grep/sed got me most of the way there. I just did a couple of manual edits and it worked fine. Thanks again for your help.

September 30th, 2006, 08:55 AM

jabs

Registered User

Join Date: Feb 2004

Posts: 19

Time spent in forums: 3 h 51 m 26 sec
Reputation Power: 0

Quote:

Originally Posted by Ehlanna

Change it to %s\n in the printf to add a newline in.

Thanks for your help! (I like that avatar too!)

October 2nd, 2006, 10:33 AM

stanleypane

Contributing User

Join Date: Oct 2003

Location: Baltimore, MD

Posts: 208

Time spent in forums: 4 Days 7 h 4 m 19 sec
Reputation Power: 21

Quote:

Originally Posted by Ehlanna

Change it to %s\n in the printf to add a newline in.

I've got the \n in there. It just has a \" just before it. He was wanting to add a quote before the newline.

October 3rd, 2006, 08:03 AM

ghostdog74

Contributing User

Join Date: Apr 2006

Posts: 100

Time spent in forums: 1 Day 20 h 48 m 7 sec
Reputation Power: 3

Quote:

Originally Posted by jabs

Here's a Python alternative, without regular expressions
Input:
1234|320|1|"Sample of data"
1234|321|2|"for DevShed"
9876|1000|1|"This sample
"
9876|1001|2|"show the bad rec
"
56574|1015|1|"Another bad one
"

Code:

>>> for lines in open("input.txt"):
... 	lines = lines.strip() #strip newlines
... 	if not lines == '"':
... 		if not lines.endswith('"'):
... 			print lines + '"'
... 		else:
... 			print lines
... 
1234|320|1|"Sample of data"
1234|321|2|"for DevShed"
9876|1000|1|"This sample"
9876|1001|2|"show the bad rec"
56574|1015|1|"Another bad one"

#10

October 4th, 2006, 08:30 AM

jabs

Registered User

Join Date: Feb 2004

Posts: 19

Time spent in forums: 3 h 51 m 26 sec
Reputation Power: 0

Quote:

Originally Posted by ghostdog74

Code:

>>> for lines in open("input.txt"):
... 	lines = lines.strip() #strip newlines
... 	if not lines == '"':
... 		if not lines.endswith('"'):
... 			print lines + '"'
... 		else:
... 			print lines
... 
1234|320|1|"Sample of data"
1234|321|2|"for DevShed"
9876|1000|1|"This sample"
9876|1001|2|"show the bad rec"
56574|1015|1|"Another bad one"

I've never used python, but I'll give this a try. Cheers!

#11

October 6th, 2006, 07:24 AM

Verletto

Registered User

Join Date: May 2006

Location: Sweden

Posts: 14

Time spent in forums: 5 h 56 m 46 sec
Reputation Power: 0

Quote:

Originally Posted by stanleypane

Code:

Method 1 - grep & sed

   grep "^[^\"]" file1 | sed -e "s/\([^\"]\)$/\1\"/g" > file2

Some of the chars can be ommited, here is a purged version:

Code:

grep -v '^"' file | sed 's/\([^"]$\)/\1"/' > file2

grep only supplies sed with one line at once therefore the 'g' option can be omitted.

That was just my 50 cents

Thread Tools	Search this Thread
Email this Page	Search this Thread: Advanced Search
Display Modes	Rate This Thread
Linear Mode Switch to Hybrid Mode Switch to Threaded Mode	Rate This Thread: