pmd/CVSROOT/syncmail

#! /usr/bin/python

# NOTE: Until SourceForge installs a modern version of Python on the cvs
# servers, this script MUST be compatible with Python 1.5.2.

"""Complicated notification for CVS checkins.

This script is used to provide email notifications of changes to the CVS
repository.  These email changes will include context diffs of the changes.
Really big diffs will be trimmed.

This script is run from a CVS loginfo file (see $CVSROOT/CVSROOT/loginfo).  To
set this up, create a loginfo entry that looks something like this:

    mymodule /path/to/this/script %%s some-email-addr@your.domain

In this example, whenever a checkin that matches `mymodule' is made, this
script is invoked, which will generate the diff containing email, and send it
to some-email-addr@your.domain.

    Note: This module used to also do repository synchronizations via
    rsync-over-ssh, but since the repository has been moved to SourceForge,
    this is no longer necessary.  The syncing functionality has been ripped
    out in the 3.0, which simplifies it considerably.  Access the 2.x versions
    to refer to this functionality.  Because of this, the script is misnamed.

It no longer makes sense to run this script from the command line.  Doing so
will only print out this usage information.

Usage:

    %(PROGRAM)s [options] <%%S> email-addr [email-addr ...]

Where options are:

    --cvsroot=<path>
    	Use <path> as the environment variable CVSROOT.  Otherwise this
    	variable must exist in the environment.

    --context=#
    -C #
        Include # lines of context around lines that differ (default: 2).

    -c
        Produce a context diff (default).

    -u
        Produce a unified diff (smaller).

    --quiet / -q
        Don't print as much status to stdout.

    --fromhost=hostname
    -f hostname
        The hostname that email messages appear to be coming from.  The From:
        header will of the outgoing message will look like user@hostname.  By
        default, hostname is the machine's fully qualified domain name.

    --help / -h
        Print this text.

The rest of the command line arguments are:

    <%%S>
        CVS %%s loginfo expansion.  When invoked by CVS, this will be a single
        string containing the directory the checkin is being made in, relative
        to $CVSROOT, followed by the list of files that are changing.  If the
        %%s in the loginfo file is %%{sVv}, context diffs for each of the
        modified files are included in any email messages that are generated.

    email-addrs
        At least one email address.
"""
__version__ = '1.2'

import os
import sys
import re
import time
import string
import getopt
import smtplib
import pwd
import socket

try:
    from socket import getfqdn
except ImportError:
    def getfqdn():
        # Python 1.5.2 :(
        hostname = socket.gethostname()
        byaddr = socket.gethostbyaddr(socket.gethostbyname(hostname))
        aliases = byaddr[1]
        aliases.insert(0, byaddr[0])
        aliases.insert(0, hostname)
        for fqdn in aliases:
            if '.' in fqdn:
                break
        else:
            fqdn = 'localhost.localdomain'
        return fqdn
    

from cStringIO import StringIO

# Which SMTP server to do we connect to?  Empty string means localhost.
MAILHOST = ''
MAILPORT = 25

# Diff trimming stuff
DIFF_HEAD_LINES = 20
DIFF_TAIL_LINES = 20
DIFF_TRUNCATE_IF_LARGER = 1000

EMPTYSTRING = ''
SPACE = ' '
DOT = '.'
COMMASPACE = ', '

PROGRAM = sys.argv[0]

BINARY_EXPLANATION_LINES = [
    "(This appears to be a binary file; contents omitted.)\n"
    ]

REVCRE = re.compile("^(NONE|[0-9.]+)$")
NOVERSION = "Couldn't generate diff; no version number found in filespec: %s"
BACKSLASH = "Couldn't generate diff: backslash in filespec's filename: %s"



def usage(code, msg=''):
    print __doc__ % globals()
    if msg:
        print msg
    sys.exit(code)



def calculate_diff(filespec, contextlines):
    file, oldrev, newrev = string.split(filespec, ',')
    # Make sure we can find a CVS version number
    if not REVCRE.match(oldrev):
        return NOVERSION % filespec
    if not REVCRE.match(newrev):
        return NOVERSION % filespec

    if string.find(file, '\\') <> -1:
        # I'm sorry, a file name that contains a backslash is just too much.
        # XXX if someone wants to figure out how to escape the backslashes in
        # a safe way to allow filenames containing backslashes, this is the
        # place to do it.  --Zooko 2002-03-17
        return BACKSLASH % filespec

    if string.find(file, "'") <> -1:
        # Those crazy users put single-quotes in their file names!  Now we
        # have to escape everything that is meaningful inside double-quotes.
        filestr = string.replace(file, '`', '\`')
        filestr = string.replace(filestr, '"', '\"')
        filestr = string.replace(filestr, '$', '\$')
        # and quote it with double-quotes.
        filestr = '"' + filestr + '"'
    else:
        # quote it with single-quotes.
        filestr = "'" + file + "'"
    if oldrev == 'NONE':
        try:
            if os.path.exists(file):
                fp = open(file)
            else:
                update_cmd = "cvs -fn update -r %s -p %s" % (newrev, filestr)
                fp = os.popen(update_cmd)
            lines = fp.readlines()
            fp.close()
            # Is this a binary file?  Let's look at the first few
            # lines to figure it out:
            for line in lines[:5]:
                for c in string.rstrip(line):
                    if c in string.whitespace:
                        continue
                    if c < ' ' or c > chr(127):
                        lines = BINARY_EXPLANATION_LINES[:]
                        break
            lines.insert(0, '--- NEW FILE: %s ---\n' % file)
        except IOError, e:
            lines = ['***** Error reading new file: ',
                     str(e), '\n***** file: ', file, ' cwd: ', os.getcwd()]
    elif newrev == 'NONE':
        lines = ['--- %s DELETED ---\n' % file]
    else:
        # This /has/ to happen in the background, otherwise we'll run into CVS
        # lock contention.  What a crock.
        if contextlines > 0:
            difftype = "-C " + str(contextlines)
        else:
            difftype = "-u"
        diffcmd = "/usr/bin/cvs -f diff -kk %s --minimal -r %s -r %s %s" \
                  % (difftype, oldrev, newrev, filestr)
        fp = os.popen(diffcmd)
        lines = fp.readlines()
        sts = fp.close()
        # ignore the error code, it always seems to be 1 :(
##        if sts:
##            return 'Error code %d occurred during diff\n' % (sts >> 8)
    if len(lines) > DIFF_TRUNCATE_IF_LARGER:
        removedlines = len(lines) - DIFF_HEAD_LINES - DIFF_TAIL_LINES
        del lines[DIFF_HEAD_LINES:-DIFF_TAIL_LINES]
        lines.insert(DIFF_HEAD_LINES,
                     '[...%d lines suppressed...]\n' % removedlines)
    return string.join(lines, '')



def blast_mail(subject, people, filestodiff, contextlines, fromhost):
    # cannot wait for child process or that will cause parent to retain cvs
    # lock for too long.  Urg!
    if not os.fork():
        # in the child
        # give up the lock you cvs thang!
        time.sleep(2)
        # Create the smtp connection to the localhost
        conn = smtplib.SMTP()
        conn.connect(MAILHOST, MAILPORT)
        user = pwd.getpwuid(os.getuid())[0]
        name = pwd.getpwuid(os.getuid())[4]
        domain = fromhost or getfqdn()
        address = '%s@%s' % (user, domain)
        s = StringIO()
        sys.stdout = s
        try:
            print '''\
From: %(name)s <%(address)s>
To: %(people)s
Subject: %(subject)s
X-Mailer: Python syncmail %(version)s <http://sf.net/projects/cvs-syncmail>
''' % {'address' : address,
       'name'    : name,
       'people'  : string.join(people, COMMASPACE),
       'subject' : subject,
       'version' : __version__,
       }
            s.write(sys.stdin.read())
            # append the diffs if available
            print
            for file in filestodiff:
                print calculate_diff(file, contextlines)
        finally:
            sys.stdout = sys.__stdout__
        resp = conn.sendmail(address, people, s.getvalue())
        conn.close()
        os._exit(0)



# scan args for options
def main():
    try:
        opts, args = getopt.getopt(
            sys.argv[1:], 'hC:cuqf:',
            ['fromhost=', 'context=', 'cvsroot=', 'help', 'quiet'])
    except getopt.error, msg:
        usage(1, msg)

    # parse the options
    contextlines = 2
    verbose = 1
    fromhost = None
    for opt, arg in opts:
        if opt in ('-h', '--help'):
            usage(0)
        elif opt == '--cvsroot':
            os.environ['CVSROOT'] = arg
        elif opt in ('-C', '--context'):
            contextlines = int(arg)
        elif opt == '-c':
            if contextlines <= 0:
                contextlines = 2
        elif opt == '-u':
            contextlines = 0
        elif opt in ('-q', '--quiet'):
            verbose = 0
        elif opt in ('-f', '--fromhost'):
            fromhost = arg

    # What follows is the specification containing the files that were
    # modified.  The argument actually must be split, with the first component
    # containing the directory the checkin is being made in, relative to
    # $CVSROOT, followed by the list of files that are changing.
    if not args:
        usage(1, 'No CVS module specified')
    subject = args[0]
    specs = string.split(args[0])
    del args[0]

    # The remaining args should be the email addresses
    if not args:
        usage(1, 'No recipients specified')

    # Now do the mail command
    people = args

    if verbose:
        print 'Mailing %s...' % string.join(people, COMMASPACE)

    if specs == ['-', 'Imported', 'sources']:
        return
    if specs[-3:] == ['-', 'New', 'directory']:
        del specs[-3:]
    elif len(specs) > 2:
        L = specs[:2]
        for s in specs[2:]:
            prev = L[-1]
            if string.count(prev, ',') < 2:
                L[-1] = "%s %s" % (prev, s)
            else:
                L.append(s)
        specs = L

    if verbose:
        print 'Generating notification message...'
    blast_mail(subject, people, specs[1:], contextlines, fromhost)
    if verbose:
        print 'Generating notification message... done.'



if __name__ == '__main__':
    main()
    sys.exit(0)