Community technical support mailing list was retired 2010 and replaced with a professional technical support team. For assistance please contact: Pre-sales Technical support via email to sales@march-hare.com.
On Wed, 11 Jul 2007 09:16:02 -0600, Glen Starrett
<glen.starrett at march-hare.com> wrote:
> Chris Purves wrote:
>> I would prefer
>> to have one e-mail represent all changes made to a module in a single
>> day.
>
> Try this:
>
> cvs -d CVSROOT rlog -d "yesterday<now" MODULE
>
> See if that produces the list you'd like to see.
>
Yes, that command was very helpful. I have incorporated it into the
script below which checks for changes that occured yesterday and sends a
message to all members of a corresponding group for module. The script
also assumes that a user's e-mail address is stored in the passwd file,
which will not be the case for most users, but is for me since I am
managing my passwd file with another program, TWiki. So, for most the
script below would need to be modifed to pull e-mails from a different
file.
Please feel free to make use of the following script.
#!/bin/bash
# This script generates a list of changes to a CVS repository
# on a module by modules bassis for all changes made yesterday.
# For each module with commits made an e-mail is sent to each
# member in a group.
#
# This script requires that sendmail is an executable
#
# This script expects that the passwd file be of the form
# username:password hash:e-mail address
# This is because my CVSNT setup uses the .htpasswd file created
# by TWiki which includes e-mail addresses. This is not the
# standard passwd configuration for CVS which has the system
# user at the end. For standard CVS setups, this script will
# need to be modified to use another method to match users
# to e-mail addresses.
#
# This script orignally written by Chris Purves
## User modifiable section
# Set locations for directores. The script will create a subdirectory
# inside the temporary directory and remove the subdirectory
# when finished. The location of the cvs repository on the
# filesystem must be entered.
tmpdir=/tmp
cvsdir=/var/lib/cvs
# list of modules seperated by spaces
modules="product_docs test"
# list of groups corresponding to modules seperated by spaces
groups="CVSProductDocsGroup CVSTestGroup"
## Main section of script. Do not modify.
# Check if temp directory exists. If it does stop the script.
# We don't want to recursively delete an existing directory.
# If all is good, create cvsmailer directory.
if [ -d $tmpdir/cvsmailer ]
then
echo "Directory $tmpdir/cvsmailer exists."
echo "Program exiting."
exit
else
mkdir $tmpdir/cvsmailer
fi
# generate begin and end dates for yesterday to be used with
# cvs rlog command
yesterday=`date -d "yesterday" +%Y-%m-%d`
begin="$yesterday 00:00"
end="$yesterday 23:59"
# Convert groups variable into group array. This may not
# be the most elegant method, but it's what I could think
# of.
i=0
for item in $groups
do
group[$i]=$item
let "i+=1"
done
# main loop
# cycle through modules
i=0
for module in $modules
do
# use cvs rlog command and write output to a file. Send errors
# to /dev/null
cvs -d :local:$cvsdir rlog -d "$begin"\<"$end" $module >
$tmpdir/cvsmailer/$module 2> /dev/null
# We only want to continue with this module if there were revisions
# made, so we check.
if grep -q "selected revisions: 0" $tmpdir/cvsmailer/$module
then
# We're really only interested in the false case, but we still
# need to do something for the true case...maybe not the best way
echo "true" > /dev/null
else
# Okay, we have revisions!
# I don't like the rlog header, so remove it.
trigger=0
# I want to use $cvsdir with sed, but the /'s cause problems, so
change them to \\/'s
sedcvsdir=`echo $cvsdir | sed 's/\//\\\\\//g'`
while read line
do
# Replace RCS file... with filename as header for section
if echo $line | grep -q "^RCS file:"
then
echo $line | sed "s/RCS file:
$sedcvsdir\/$module\/\(.*\),v/********** \1 **********/" >>
$tmpdir/cvsmailer/tempfile
fi
# Only write lines where trigger = 1
if [ $trigger -eq 1 ]
then
echo "$line" >> $tmpdir/cvsmailer/tempfile
fi
# The line beginning "description" marks the end of the log header
# Once we see that we can set trigger to 1
if [ "$line" = "description:" ]
then
trigger=1
fi
# The line beginning "=====" marks the end of a section. Reset
# trigger to 0 so that next header is not written to output.
if echo $line | grep -q ======
then
trigger=0
fi
done < $tmpdir/cvsmailer/$module
mv $tmpdir/cvsmailer/tempfile $tmpdir/cvsmailer/$module
# Pull users from the group that corresponds to this module.
names=`grep "${group[$i]}" $cvsdir/CVSROOT/group | sed 's/^.*: //'`
# For each user, find their e-mail address from the passwd file
# and store in a variable $address
for name in $names
do
email=`grep $name $cvsdir/CVSROOT/passwd | sed 's/.*://'`
address="$address, $email"
done
address=`echo $address | sed 's/^, //'`
# Create an e-mail template and append the log
# The first line overwrites the file in case there are more than
# one module with changes. After that append.
echo "From: \"CVS Mailer\" <pchris at aims-bio.com>" >
$tmpdir/cvsmailer/cvsmail
echo "To: $address" >> $tmpdir/cvsmailer/cvsmail
echo "Subject: CVS Commit to $module" >> $tmpdir/cvsmailer/cvsmail
echo "" >> $tmpdir/cvsmailer/cvsmail
echo "The following changes have been made to the $module module." >>
$tmpdir/cvsmailer/cvsmail
echo "" >> $tmpdir/cvsmailer/cvsmail
cat $tmpdir/cvsmailer/$module >> $tmpdir/cvsmailer/cvsmail
# Use sendmail command to send the e-mail.
sendmail -t < $tmpdir/cvsmailer/cvsmail
fi
# Increment to correctly index the group array.
let "i+=1"
done
# Remove the created directory inside the tmpdir. Since we checked
# that this directory did not exist at the start, it should be safe.
rm -r $tmpdir/cvsmailer
--
Chris Purves