BashScripts.org

It is currently Wed Apr 16, 2014 3:52 pm

extract text using sed

Board index » The Scripts » Requests

All times are UTC - 6 hours

Page 1 of 2

[ 31 posts ]

Go to page 1, 2 Next

Print view

Previous topic | Next topic

Author

Message

BrianUK

Posted: Sun Feb 06, 2011 2:42 pm

Joined: Fri May 16, 2008 4:58 am
Posts: 94

I'm using curl to get the below output but only want to get the output from in between the <LocalityName></LocalityName> tags. I can wittle it down using nawk 'NR==10' but would prefer using sed or similar, can anyone help?

Code:

<?xml version="1.0" encoding="UTF-8" ?>
<kml xmlns=""><Response>
  <name></name>
  <Status>
    <code>200</code>
    <request>geocode</request>
  </Status>
  <Placemark id="p1">
    <address></address>
    <AddressDetails Accuracy="5" xmlns="urn:oasis:names:tc:ciq:xsdschema:xAL:2.0"><Country><CountryNameCode></CountryNameCode><CountryName>UK</CountryName><Locality><LocalityName>EXTRACT</LocalityName><PostalCode><PostalCodeNumber></PostalCodeNumber></PostalCode></Locality></Country></AddressDetails>
    <ExtendedData>
      <LatLonBox north="0000000000" south="00000000" east="00000000" west="000000" />
    </ExtendedData>
    <Point><coordinates>00000000</coordinates></Point>
  </Placemark>
</Response></kml>

Top

scilec

Posted: Sun Feb 06, 2011 4:29 pm

Joined: Sat Jan 16, 2010 5:53 pm
Posts: 1

Hi,
I'm sure there's a more elegant way to do this but this is the first thing that popped into my head and it should get the job done for now. In a nutshell, the grep statement below grabs only the lines that contain "<LocalityName>" and then that output is piped into sed, which deletes everything from the beginning of the line (^) to "<LocalityName>" and then from "</LocalityName>" (in which the "/" is escaped by preceding it with a "\") to the end of the line.

$ grep "<LocalityName>" filename.xml | sed -r 's/^.*<LocalityName>//g; s/<\/LocalityName>.*$//g'

Top

Patsie

Posted: Sun Feb 06, 2011 11:36 pm

Joined: Sun Jun 27, 2010 12:57 am
Posts: 192

Print only what is between the 2 named tags:

Code:

sed -n 's/.*<LocalityName>$.*$<\/LocalityName>.*/\1/p'

Top

BrianUK

Posted: Mon Feb 07, 2011 4:14 am

Joined: Fri May 16, 2008 4:58 am
Posts: 94

Patsie wrote:

Print only what is between the 2 named tags:

Code:

sed -n 's/.*<LocalityName>$.*$<\/LocalityName>.*/\1/p'

works a treat thanks,

i'm extracting gps info using

Code:

sed -rn '/<trkpt lat="/{s%.* lat="([^"]+)" lon="([^"]+)">.*%http://maps.google.com/maps?q=\1,\2%;p}'

which returns http://maps.google.com/maps?q=0.0000000,-0.000000

after the last 0 I'd like it to add &output=xml so can anybody help me teak the above code to do this?

using the returned url i'd like to be able to extrack the Locality info ie

Code:

curl -s "URL" sed -n 's/.*<LocalityName>$.*$<\/LocalityName>.*/\1/p'

and finally put it into a script ie

GPS="blah blah"
echo "$GPS"

Top

BrianUK

Posted: Fri Feb 11, 2011 10:29 am

Joined: Fri May 16, 2008 4:58 am
Posts: 94

anyone?

Top

jeo

Posted: Fri Feb 11, 2011 1:00 pm

Moderator

Joined: Wed May 03, 2006 2:05 pm
Posts: 242

Hi BrianUK!

I think I might be missing something here. Could we see some sample data for the file or output you're using to generate the URL?

Thanks!
-Jeo

Top

BrianUK

Posted: Sat Feb 12, 2011 3:22 pm

Joined: Fri May 16, 2008 4:58 am
Posts: 94

In bash using my gps location software along with sed I'm returned

http://maps.google.com/maps/maps?q=00.0000,-0.00000

at the end of this url i'd like to add &output=xml

From the output of that url i'd like to run this command

Code:

curl -s "URL" sed -n 's/.*<LocalityName>$.*$<\/LocalityName>.*/\1/p'

which would give me for example 'London'

Top

Patsie

Posted: Sun Feb 13, 2011 1:12 am

Joined: Sun Jun 27, 2010 12:57 am
Posts: 192

So, lets assume you google-maps URL is stored in $URL
you just run:

Code:

curl -s "${URL}&output=xml" | sed -n 's/.*<LocalityName>\(.*\)<\/LocalityName>.*/\1/p'

And you should be done, right?

Top

BrianUK

Posted: Sun Feb 13, 2011 5:56 am

Joined: Fri May 16, 2008 4:58 am
Posts: 94

getting close

Code:

#!/bin/bash
URL=gps | sed -rn '/<trkpt lat="/{s%.* lat="([^"]+)" lon="([^"]+)">.*%http://maps.google.com/maps/map?q=\1,\2%;p}'

LOCATION=curl -s "${URL}&output=xml" | sed -n 's/.*<LocalityName>\(.*\)<\/LocalityName>.*/\1/p'

while true
do

echo "$LOCATION"

Would this be the right way to do it?

./gps: line 4: -s: command not found

Top

Patsie

Posted: Sun Feb 13, 2011 7:40 am

Joined: Sun Jun 27, 2010 12:57 am
Posts: 192

If you want to put the output of a command in a variable, use something line this:

Code:

var=$(command -args)

The way you do it, var=command -args, doesn't work.

Last edited by Patsie on Sun Feb 13, 2011 7:42 am, edited 1 time in total.

Top

Patsie

Posted: Sun Feb 13, 2011 7:41 am

Joined: Sun Jun 27, 2010 12:57 am
Posts: 192

double post, sorry.

Top

BrianUK

Posted: Sun Feb 13, 2011 8:08 am

Joined: Fri May 16, 2008 4:58 am
Posts: 94

now i'm confused ha, is this for the URL or location bit? could you place it into the script for me?

Top

jeo

Posted: Mon Feb 14, 2011 8:06 am

Moderator

Joined: Wed May 03, 2006 2:05 pm
Posts: 242

What Patsie is saying is that in order to store the output of a command in a variable, you have to surround the command in $() or ``

So for our example:

Code:

# In your script, you're doing this, which will not work:
VARIABLE=commands to execute here

# The proper way looks more like this:
VARIABLE=$(commands to execute here)

# Here's an alternative way, with `back-ticks`
# But the one above is usually preferred
VARIABLE=`commands to execute here`

# To use YOUR script as an example:
URL=$(gps | sed -rn '/<trkpt lat="/{s%.* lat="([^"]+)" lon="([^"]+)">.*%http://maps.google.com/maps/map?q=\1,\2%;p}')
LOCATION=$(curl -s "${URL}&output=xml" | sed -n 's/.*<LocalityName>\(.*\)<\/LocalityName>.*/\1/p')

Everything inside "$()" will be executed, and the ouptut will be stored in the variable you specified.

I hope this helps!
-J

Top

BrianUK

Posted: Mon Feb 14, 2011 2:13 pm

Joined: Fri May 16, 2008 4:58 am
Posts: 94

Ahh I see what you mean I remember i've previously used back ticks and forgot them in my example, i'm tring your verison with

Code:

echo "$LOCATION"

at the end but not getting any output?

Top

BrianUK

Posted: Thu Feb 17, 2011 7:42 am

Joined: Fri May 16, 2008 4:58 am
Posts: 94

Top

jeo

Posted: Thu Feb 17, 2011 10:43 am

Moderator

Joined: Wed May 03, 2006 2:05 pm
Posts: 242

Hi BrianUK!

Try checking the output of your curl command by itself, without running it through sed. In my script, I get an error:

The requested URL <code>/maps/map?q=37.646182,-115.750827&output=xml</code> was not found on this server.

BUT if I change the URL a bit... by replacing /maps/map with maps, I get the source of the map page, but it doesn't seem to understand the output=xml parameter.

It looks like you may have intended to use the google maps api?

Top

jeo

Posted: Thu Feb 17, 2011 11:29 am

Moderator

Joined: Wed May 03, 2006 2:05 pm
Posts: 242

This script gives us some XML output, but it's not formatted the way your sed expression is looking for it.

Instead of <LocalityName> we have something like:

Code:

<address_component>
    <long_name></long_name>
    <short_name></short_name>
    <type>locality</type>
</address_component>

Here's what I've got so far:

Code:

#!/bin/bash

# I don't know what the gps command is/does so I
# created a dummy file with data that matches the regex
gps () {
    cat tmp.txt
}

URL=$(gps | sed -rn '/<trkpt lat="/{s%.* lat="([^"]+)" lon="([^"]+)">.*%http://maps.googleapis.com/maps/api/geocode/xml?latlng=\1,\2\&sensor=false%;p}')
OUTPUT=$(curl -s "${URL}" | grep -B2 locality)

echo "$URL"
echo "$OUTPUT"

Top

BrianUK

Posted: Thu Feb 17, 2011 1:32 pm

Joined: Fri May 16, 2008 4:58 am
Posts: 94

ok for what ever reason my 'gps' software doesn't work being called via a script. This works fine in command line

Code:

gps | sed -rn '/<trkpt lat="/{s%.* lat="([^"]+)" lon="([^"]+)">.*%http://maps.google.com/maps?q=\1,\2%;p}'

but in a bash script using just

Code:

#!/bin/bash

URL=`gps | sed -rn '/<trkpt lat="/{s%.* lat="([^"]+)" lon="([^"]+)">.*%http://maps.google.com/maps?q=\1,\2%;p}'`

echo "$URL"

it doesn't do anything so looks like i'll have to try it a different way.

Ah just run using -x and it appears to be running gps as one command and sed as another

++ gps
++ sed -rn '/<trkpt lat="/{s%.* lat="([^"]+)" lon="([^"]+)">.*%http://maps.google.com/maps?q=\1,\2%;p}'

any idea how to fix this?

Top

jeo

Posted: Thu Feb 17, 2011 1:46 pm

Moderator

Joined: Wed May 03, 2006 2:05 pm
Posts: 242

Does your gps command require any parameters? That bash -x output looks normal (well... the two lines that you pasted anyway). Here are the first few lines of mine:

Code:

$ bash -x tmp.sh 
++ gps
++ cat tmp.txt
++ sed -rn '/<trkpt lat="/{s%.* lat="([^"]+)" lon="([^"]+)">.*%http://maps.googleapis.com/maps/api/geocode/xml?latlng=\1,\2\&sensor=false%;p}'
+ URL='http://maps.googleapis.com/maps/api/geocode/xml?latlng=37.646182,-115.750827&sensor=false'

Does your bash -x run get as far as saying '+ URL=...'?

Top

BrianUK

Posted: Thu Feb 17, 2011 2:03 pm

Joined: Fri May 16, 2008 4:58 am
Posts: 94

no it's doesn't need any parameters

no I only get the 2 lines I posted returned

works ok from the command line

Code:

gps | sed -rn '/<trkpt lat="/{s%.* lat="([^"]+)" lon="([^"]+)">.*%http://maps.googleapis.com/maps/api/geocode/xml?latlng=\1,\2\&sensor=false%;p}'

http://maps.googleapis.com/maps/api/geocode/xml?latlng=00.000000,-0.00000&sensor=false

Top

jeo

Posted: Thu Feb 17, 2011 2:42 pm

Moderator

Joined: Wed May 03, 2006 2:05 pm
Posts: 242

Interesting... Here's a simpler test... What happens if you just run 'gps' from a script, and don't try to parse the output?

Code:

echo -e '#!/bin/bash\ngps' > mytest.sh
chmod +x mytest.sh
./mytest.sh

Top

BrianUK

Posted: Thu Feb 17, 2011 2:58 pm

Joined: Fri May 16, 2008 4:58 am
Posts: 94

works fine, even if I do

Code:

#!/bin/bash

gps | sed -rn '/<trkpt lat="/{s%.* lat="([^"]+)" lon="([^"]+)">.*%http://maps.googleapis.com/maps/api/geocode/xml?latlng=\1,\2\&sensor=false%;p}'

but add a but more and a couple of ticks

Code:

#!/bin/bash

URL=`gps | sed -rn '/<trkpt lat="/{s%.* lat="([^"]+)" lon="([^"]+)">.*%http://maps.google.com/maps?q=\1,\2%;p}'`

echo "$URL"

and I get nothing

Top

jeo

Posted: Thu Feb 17, 2011 3:26 pm

Moderator

Joined: Wed May 03, 2006 2:05 pm
Posts: 242

There's only one more thing I can think of to try. Instead of `backticks` try using enclosing it in $(here) like in our examples above.

Top

BrianUK

Posted: Thu Feb 17, 2011 3:58 pm

Joined: Fri May 16, 2008 4:58 am
Posts: 94

Code:

./location5: line 5: syntax error near unexpected token `|'
./location5: line 5: `URL=(gps | sed -rn '/<trkpt lat="/{s%.* lat="([^"]+)" lon="([^"]+)">.*%http://maps.googleapis.com/maps/api/geocode/xml?latlng=\1,\2\&sensor=false%;p}')'

Top

jeo

Posted: Fri Feb 18, 2011 10:31 am

Moderator

Joined: Wed May 03, 2006 2:05 pm
Posts: 242

You're missing a "$"

It should look like one of the many examples above:

Code:

## Generic example (thanks Patsie!):
var=$(command -args)

## Specific example:
URL=$(gps | sed -rn '/<trkpt lat="/{s%.* lat="([^"]+)" lon="([^"]+)">.*%http://maps.googleapis.com/maps/api/geocode/xml?latlng=\1,\2\&sensor=false%;p}')

I hope this helps!

Top

Page 1 of 2

[ 31 posts ]

Go to page 1, 2 Next

Board index » The Scripts » Requests

All times are UTC - 6 hours

Who is online

Users browsing this forum: No registered users and 6 guests

You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot post attachments in this forum

BashScripts | Promote Your Page Too

extract text using sed

Who is online