pmd/CVSROOT/syncmail

#! /usr/bin/python

# Copyright (c) 2002, 2003, Barry Warsaw, Fred Drake, and contributors
# All rights reserved.
# See the accompanying LICENSE file for details.

# NOTE: SourceForge currently runs Python 2.2.3, so we need to remain
# compatible with the Python 2.2 line.

"""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).

    -m hostname
    --mailhost hostname
        The hostname of an available SMTP server.  The default is
        'localhost'.

    -u
        Produce a unified diff (smaller).

    -S TEXT
    --subject-prefix=TEXT
        Prepend TEXT to the email subject line.

    -R ADDR
    --reply-to=ADDR
      Add a "Reply-To: ADDR" header to the email message.

    --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 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__ = '1.3'

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

from cStringIO import StringIO

# Which SMTP server to do we connect to?
MAILHOST = 'localhost'
MAILPORT = 25

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

COMMASPACE = ', '

PROGRAM = sys.argv[0]

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

NOVERSION = "Couldn't generate diff; no version number found for file: %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(entry, contextlines):
    file = entry.name
    oldrev = entry.revision
    newrev = entry.new_revision

    # Make sure we can find a CVS version number
    if oldrev is None and newrev is None:
        return NOVERSION % file

    if file.find("'") <> -1:
        # Those crazy users put single-quotes in their file names!  Now we
        # have to escape everything that is meaningful inside double-quotes.
        filestr = filestr.replace('\\', '\\\\')
        filestr = filestr.replace('`', '\`')
        filestr = filestr.replace('"', '\"')
        filestr = filestr.replace('$', '\$')
        # and quote it with double-quotes.
        filestr = '"' + filestr + '"'
    else:
        # quote it with single-quotes.
        filestr = "'" + file + "'"
    if oldrev is None:
        # File is being added.
        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 line.rstrip():
                    if c.isspace():
                        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 is None:
        lines = ['--- %s DELETED ---\n' % file]
    else:
        # File has been changed.
        # 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()
        # ignore the error code, it always seems to be 1 :(
        fp.close()
    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 ''.join(lines)



rfc822_specials_re = re.compile(r'[\(\)\<\>\@\,\;\:\\\"\.\[\]]')

def quotename(name):
    if name and rfc822_specials_re.search(name):
        return '"%s"' % name.replace('"', '\\"')
    else:
        return name



def blast_mail(subject, people, entries, contextlines, fromhost, replyto):
    # 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)
        pwinfo = pwd.getpwuid(os.getuid())
        user = pwinfo[0]
        name = pwinfo[4].split(',')[0]
        domain = fromhost or socket.getfqdn()
        address = '%s@%s' % (user, domain)
        s = StringIO()
        sys.stdout = s
        datestamp = time.strftime('%a, %d %b %Y %H:%M:%S +0000',
                                  time.gmtime(time.time()))
        try:
            vars = {'address' : address,
                    'name'    : quotename(name),
                    'people'  : COMMASPACE.join(people),
                    'subject' : subject,
                    'version' : __version__,
                    'date'    : datestamp,
                    }
            print '''\
From: %(name)s <%(address)s>
To: %(people)s''' % vars
            if replyto:
                print 'Reply-To: %s' % replyto
            print '''\
Subject: %(subject)s
Date: %(date)s
X-Mailer: Python syncmail %(version)s <http://sf.net/projects/cvs-syncmail>
''' % vars
            s.write(sys.stdin.read())
            # append the diffs if available
            print
            for entry in entries:
                print calculate_diff(entry, contextlines)
        finally:
            sys.stdout = sys.__stdout__
        resp = conn.sendmail(address, people, s.getvalue())
        conn.close()
        os._exit(0)



class CVSEntry:
    def __init__(self, name, revision, timestamp, conflict, options, tagdate):
        self.name = name
        self.revision = revision
        self.timestamp = timestamp
        self.conflict = conflict
        self.options = options
        self.tagdate = tagdate

def get_entry(prefix, mapping, line, filename):
    line = line.strip()
    parts = line.split("/")
    _, name, revision, timestamp, options, tagdate = parts
    key = namekey(prefix, name)
    try:
        entry = mapping[key]
    except KeyError:
        if revision == "0":
            revision = None
        if timestamp.find("+") != -1:
            timestamp, conflict = tuple(timestamp.split("+"))
        else:
            conflict = None
        entry = CVSEntry(key, revision, timestamp, conflict,
                         options, tagdate)
        mapping[key] = entry
    return entry

def namekey(prefix, name):
    if prefix:
        return os.path.join(prefix, name)
    else:
        return name

def load_change_info(prefix=None):
    if prefix is not None:
        entries_fn = os.path.join(prefix, "CVS", "Entries")
    else:
        entries_fn = os.path.join("CVS", "Entries")
    entries_log_fn = entries_fn + ".Log"
    mapping = {}
    f = open(entries_fn)
    while 1:
        line = f.readline()
        if not line:
            break
##        if line.strip() == "D":
##            continue
        # we could recurse down subdirs, except the Entries.Log files
        # we need haven't been written to the subdirs yet, so it
        # doesn't do us any good
##        if line[0] == "D":
##            name = line.split("/")[1]
##            dirname = namekey(prefix, name)
##            if os.path.isdir(dirname):
##                m = load_change_info(dirname)
##                mapping.update(m)
        if line[0] == "/":
            # normal file
            get_entry(prefix, mapping, line, entries_fn)
        # else: bogus Entries line
    f.close()
    if os.path.isfile(entries_log_fn):
        f = open(entries_log_fn)
        while 1:
            line = f.readline()
            if not line:
                break
            if line[1:2] != ' ':
                # really old version of CVS
                break
            entry = get_entry(prefix, mapping, line[2:], entries_log_fn)
            parts = line.split("/")[1:]
            if line[0] == "A":
                # adding a file
                entry.new_revision = parts[1]
            elif line[0] == "R":
                # removing a file
                entry.new_revision = None
        f.close()
    for entry in mapping.values():
        if not hasattr(entry, "new_revision"):
            print 'confused about file', entry.name, '-- ignoring'
            del mapping[entry.name]
    return mapping

def load_branch_name():
    tag_fn = os.path.join("CVS", "Tag")
    if os.path.isfile(tag_fn):
        f = open(tag_fn)
        line = f.readline().strip()
        f.close()
        if line[:1] == "T":
            return line[1:]
    return None

# scan args for options
def main():
    # XXX Should really move all the options to an object, just to
    # avoid threading so many positional args through everything.
    try:
        opts, args = getopt.getopt(
            sys.argv[1:], 'hC:cuS:R:qf:m:',
            ['fromhost=', 'context=', 'cvsroot=', 'mailhost=',
             'subject-prefix=', 'reply-to=',
             'help', 'quiet'])
    except getopt.error, msg:
        usage(1, msg)

    # parse the options
    contextlines = 2
    verbose = 1
    subject_prefix = ""
    replyto = None
    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 ('-S', '--subject-prefix'):
            subject_prefix = arg
        elif opt in ('-R', '--reply-to'):
            replyto = arg
        elif opt in ('-q', '--quiet'):
            verbose = 0
        elif opt in ('-f', '--fromhost'):
            fromhost = arg
        elif opt in ('-m', '--mailhost'):
            global MAILHOST
            MAILHOST = 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 = subject_prefix + args[0]
    specs = args[0].split()
    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 specs[-3:] == ['-', 'Imported', 'sources']:
        print 'Not sending email for imported sources.'
        return

    branch = load_branch_name()
    changes = load_change_info()

    if verbose:
        print 'Mailing %s...' % COMMASPACE.join(people)
        print 'Generating notification message...'
    blast_mail(subject, people, changes.values(),
               contextlines, fromhost, replyto)
    if verbose:
        print 'Generating notification message... done.'


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