Shell Programming and Scripting

Hello experts!

I have a file1 with the following format (yr,day, month, hour,minute):

Code:

201201132435
201202141210
201304132030
201410100110
...

What i want to do is to assign variables and then use them in the curl command to download the text of each event from a web page.

What I have done so far is:

Code:

#!/bin/sh
#
yr=`awk '{$1=substr($1, 6, 2)}1'  file1`
mo=`awk '{$1=substr($1, 8, 2)}1'  file1`
da=`awk '{$1=substr($1, 10, 2)}1'  file1`
ho=`awk '{$1=substr($1, 12, 2)}1'  file1`
mi=`awk '{$1=substr($1, 14, 2)}1' file1 `
#
curl -L http://webpage/${yr}/${mo}/track${yr}${mo}${da}${ho}${mi}$_infos.html

but it seems that the variables are not evolving gradually all together but separately

Code:

yr=`awk '{$1=substr($1, 6, 2)}1'  file1`

yr will not hold just one year, but every year for each line in file1. The same for each of your variables mo, da, ho and mi. Nonetheless, the characters extracted are off: substr($1, 6, 2) doesn't return the representation of a year.

Code:

$ awk '{$1=substr($1, 6, 2)}1' file1 
11
21
41
01

This will do:

Code:

$ awk '{$1=substr($1, 1, 4)}1' file1
2012
2012
2013
2014

Perhaps, try:

Code:

while read d; do
    yr=${d:0:4}
    mo=${d:4:2}
    da=${d:6:2}
    ho=${d:8:2}
    mi=${d:10:2}
    curl -L http://webpage/${yr}/${mo}/track${yr}${mo}${da}${ho}${mi}_infos.html
    # if yr is supposed to be yyyy
    # curl -L http://webpage/${yr}/${mo}/track${d}_infos.html
    
done < file1

If yr is supposed to be represented with only the last two digits, replace yr=${d:0:4} to yr=${d:2:2}

Of course, the whole process can be reduced to:

Code:

while read d; do
    #curl -L http://webpage/${d:0:4}/${d:4:2}/track${d}_infos.html
    # or
    #curl -L http://webpage/${d:2:2}/${d:4:2}/track${d:2}_infos.html
done < file1

What Aia suggested is fine if you're using a system where /bin/sh is not a pure Bourne shell (as it is on Solaris systems) or a 1988 Korn shell (as it is on many AIX systems). But, in addition to what Aia already commented on, there is nothing in your script that sets the shell variable _infos, so, unless that variable is being inherited from an exported variable in the environment of your script, the $_infos in your invocations of curl will be replaced by an empty string by the shell each time it invokes curl.

When posting questions to these forums, it always helps us provide answers that will work better in your environment if you tell us what operating system and shell you're using when you post questions. And, if you're using shell variables that aren't defined in your script, please tell us where those variables are being initialized.

If you are using a Solaris system or an AIX system or an HP/UX system, you're likely to get a bunch of syntax errors from Aia's suggestion, and we'll need to know which OS you're using, whether we can use a shell other than /bin/sh on your system (such as /usr/xpg4/bin/sh on a Solaris system or /bin/ksh93 on AIX) so we'll know if we can just use a different shell or if we need to convert the shell substring expansion into a single awk script (if you're stuck with a pure Bourne shell) or into POSIX conforming variable expansions in other shells.

Thank you Aia it perfectly worked! no words! Don Cragun thank you for the suggestion. The Aia's code worked on my opensuse 13.2 system because "_infos" is not a variable but plain text.

With a recent shell, this might work:

Code:

while { read -n4 A; read -n2 B; read -n2 C; read -n2 D; read -n2 E; read REST; }; do echo $A $B $C $D $E; done < file
2012 01 13 24 35
2012 02 14 12 10
2013 04 13 20 30
2014 10 10 01 10

It looks like Aia guessed correctly by ignoring the last dollar-sign in your original post:

Code:

curl -L http://webpage/${yr}/${mo}/track${yr}${mo}${da}${ho}${mi}$_infos.html

When the shell processes $_infos.html, it treats _infos as a variable name, and, if it is unset, expands $_infos.html to the string .html; not to the string _infos.html.

my fault the "$_infos" is wrong