#!/var/tmp/py/bin/python2.3

# Adapted by bdolicki for openacs.org
# Changes to http://dev.zope.org/CVS/postcommit_actions :
#   001  The original file forced python2.1.  We are using python2.3 and it works
#   fine.
#   002  Set some variables
#   003  We want to put something like username@openacs.org as the sender email.
#        (too bad we don't have LDAP, actually)

"""Apply checkin actions dictated by traffic_table.py, as checkins occur.

Special provisions to fork into the background, disconnected from tty, so
the user and (more critically) CVS locks aren't held just pending completion 
of notification and other potentially meandering actions.

Options:

 --verbose - operate in verbose mode, with output to OUTPUT_LOG in the
             detached copy, and to stdout before the detach (and forever if 
             --wait specified)

 --dryrun - do not actually do any of the activities

 --wait - do not fork into the background (and leave output in stdout/stderr)
          This is for internal use by the script when forking.

 --msgfile MSGFILENM - internal - get log message from MSGFILENM (and delete)

We expect to be invoked from loginfo with %{CVSROOT} %{sVv}: CVS expands
the loginfo %{sVv} into a single command-line token (bizarrely enough)
containing multiple, space-delimited parts:

  - The first part is the path of repository directory

  - Then comes a comma-concatenated string for each file being checked in,
    consisting of:

    - The file name,
    - The old version number, then
    - the new one, like so:

      filename,1.52,1.53

\(See the CVS loginfo docs for more format-string details.)

The actions are configured in traffic_table.py.  The only remaining relevant
one is:

 - email checkin messages to designated recipients

\(Versions of the script that supported automatic mirroring was removed
beginning at version 1.123.)"""

import sys, os, tempfile, time, pwd
SCRIPT_DIR = os.path.abspath(os.path.split(sys.argv[0])[0])
os.chdir(SCRIPT_DIR)                # Valid dir, having other local scripts.
import string
import getopt
import smtplib
import socket, whrandom, getpass

CVSROOT = os.path.split(SCRIPT_DIR)[0]
CVSROOT_ABS = os.path.abspath('..')
# 002
#USE_LDAP = 1
USE_LDAP = 0
LDAP_PORT = 389
LDAP_HOST = "your.ldap.host"
# 002 TODO is it OK to use this address?
#OFFICIAL_SENDER = "cvs-admin@your.domain"
OFFICIAL_SENDER = "openacs-cvs-list-admin@willfork.com"

# 003
MY_DOMAIN = openacs.org

# Ceiling on number of lines per file report (diff or "added"), at which point
# we go to excerpts from the beginning and end:
FILE_LINES_LIMIT = 1500
# Number of lines in beginning and end excerpts when FILE_LINES_LIMIT is hit:
FILE_EXCERPTS_LINES = 200

CUSTOM_TRAFFIC_TABLE = "%s/custom_traffic_table.py" % CVSROOT

import re

import traffic_table
import adjustlinks

# Hook up with an outboard file.
if os.path.exists(CUSTOM_TRAFFIC_TABLE):
    execfile(CUSTOM_TRAFFIC_TABLE)

# OUTPUT_LOG must be writable by anyone whose checkins will invoke this
# script in order for their logging to happen.
OUTPUT_LOG = "%s/cvs-postcommit_actions.log" % CVSROOT

# Set CVSMASTER to the email address of someone managing your CVS mirroring.
# Notices about any caught malfunctions will be sent to this address.
# 002 TODO is it OK to use this address?
#CVSMASTER = "cvs-admin@zope.com"
CVSMASTER = "openacs-cvs-list-admin@willfork.com"

# You have to plug in the recipients email address via '%' string formatting.
MAIL_CMD = "/usr/lib/sendmail -t -f %s"

WAIT = 0                                # Fork unless --wait is requested.
DRYRUN = 0                              # '--dryrun' option
VERBOSE = 0                             # '--verbose' option default value
SCRIPT = "postcommit_actions"           # Will b set to argv[0], when obtained.

sVv_re = re.compile("(.*),([^,]+),([^,]+)")

def main(args):
    """Grok the args and the traffic_table and process accordingly."""

    global SCRIPT, DRYRUN, VERBOSE, WAIT

    orig_args = args[:]

    SCRIPT = args[0]
    try:
        opts, args = getopt.getopt(args[1:], "", ["verbose", "dryrun",
                                                  "wait", "msgfile=",
                                                  ])
    except getopt.error, err:
        complain("%s\n", err)
        raise SystemExit, 1

    msgfilenm = ""

    for opt, val in opts:
        if opt == "--verbose":
            VERBOSE = 1
            complain("%s: VERBOSE\n", SCRIPT)
        elif opt == "--dryrun":
            DRYRUN = 1
            complain("%s: DRYRUN\n", SCRIPT)
        elif opt == "--wait":
            WAIT = 1
            complain("%s: WAIT/Foreground\n", SCRIPT)
        elif opt == "--msgfile":
            # This is internal, for the script to pass itself a file.
            msgfilenm = val
        else:
            complain("Unrecognized option '%s'\n", opt)
            raise SystemExit, 1
    if VERBOSE:
        complain("Initial args: %s '%s'\n",
                 string.join(orig_args[:-1], " "),
                 orig_args[-1])
        complain("Working dir: %s\n" % safe_getcwd('/tmp'))

    if len(args) != 1:
        usage(); raise SystemExit, 1

    doing_directory = 0
    no_files = 1

    # The %{sVv} args are passed in as a single token - have to split them.
    split_args = string.split(args[0])
    repo = split_args[0]

    if len(split_args) == 0:
        subjs = ["<no files>", "", ""]
    elif len(split_args) == 1:
        subjs = split_args + ["", ""]
    elif split_args[1] == "-":
        subjs = split_args
        doing_directory = 1
    else:
        no_files = 0
        subjs = map(grok_file, split_args[1:])

    if VERBOSE:
        complain("CWD: %s, Repo: %s, ",
                 safe_getcwd('/tmp'), repo)
        complain("Subjects: %s\n", subjs)

    if not WAIT:
        detach(orig_args)

    entries = find_entries(repo)

    for entry in entries:
        wasVerbose = VERBOSE
        VERBOSE = entry.get('verbose', VERBOSE)
        selector_path = entry['path']
        addrs = entry.get('addrs')
        specials = entry.get('specials')

        if addrs:
            do_mail(repo, addrs, subjs, msgfilenm=msgfilenm,
                    doing_directory=doing_directory, no_files=no_files)

        if specials:
            subj_names = map(lambda x: x[0], subjs)
            for trigger, action in specials:
                if trigger in subj_names:
                    do_special(trigger, action, addrs)

        VERBOSE = wasVerbose

    if VERBOSE:
        complain("** Done **\n\n")

    if failures:
        handle_failures(orig_args)

    if msgfilenm:
        os.unlink(msgfilenm)

def detach(args):
    """Fork us into the background, with stdout and stderr to OUTPUT_LOG.

    We have to disconnect std io (in, out, and err) and run the program in
    the background - which we do using some shell wizardry."""

    doctored_args = [args[0]]
    doctored_args.append("--wait")
    tempfile.mktemp()           # Throw one away, to get initial template.
    tempfile.tempdir = "/tmp"
##    template = (tempfile.template or
##                "@%s.%d" % (os.getpid(), whrandom.randint(1000000, 9999999)))
##    tempfile.template = "cvs-log" + template
    msgfile = open(tempfile.mktemp(suffix='.cvslog'), 'w')
    msgfile.write(sys.stdin.read())
    msgfile.close()
    msgfilenm = msgfile.name
    doctored_args.append("--msgfile")
    doctored_args.append(msgfilenm)
    for i in args[1:]:
        doctored_args.append('"%s"' % i)
    cmd = (("( exec 1>>%s 2>&1; "
            + ("export CVSROOT=%s; " % CVSROOT)
            + string.join(doctored_args, " ") +  " &)")
           % OUTPUT_LOG)
    
    if VERBOSE:
        complain("Re-executing detached in %s, cmd:\n\t%s\n",
                 safe_getcwd('/tmp'), cmd)

    os.system(cmd)
    loosen_file(OUTPUT_LOG)

    os._exit(0)

def find_entries(repo=None):
    """Return dictionary of traffic_table entries qualified by repo regexp.

    Iff no entries match, we return the catchall entry - the (last) one with
    path == None."""
    
    entries = []
    catchall = None
    linkmgr = adjustlinks.LinkManager()
    gotaddrs = []

    containers = linkmgr.all_containers(repo)
    for it in traffic_table.get_table():
        if it['path'] == None:
            # Retain the catchall entry in case no regular ones qualify.
            catchall = it
        else:
            # Obtain qualifying candidates:
            for candidate in containers:

                if (re.match(it['path'], candidate)
                    and it.get('addrs') not in gotaddrs):
                    entries.append(it)
                    gotaddrs.append(it.get('addrs'))

    if entries:
        if VERBOSE > 1:
            complain("find_entries: repo: %s, containers: %s\n entries: %s\n",
                     repo, containers, entries)
    elif catchall:
        entries.append(catchall)
        if VERBOSE > 1:
            complain("No matches, so using catchall:\n  %s\n", entries)
    elif VERBOSE > 1:
        complain("No matches, no catchall - no actions\n")

    for e in entries:
        if e.has_key('addrs') and (type(e['addrs']) == type("")):
            # Be lenient - listify string args.
            e['addrs'] = [e['addrs']]

    return entries

def do_mail(repo, addrs, subjs,
            msgfilenm, doing_directory=0, no_files=0):
    """Send notice about checkin to addresses dictated by traffic table.

    We include a diff."""
    if VERBOSE:
        complain("Notice to %s\n for %s / %s\n", addrs, repo, subjs)
    # The message contents are on stdin, just _yearning_ to be sent...-)
    subject = "CVS: %s " % repo

    diff_msg = ''

    if doing_directory or no_files:
        subject = subject + string.join(subjs, " ")
        if subjs[2] == 'New':
            new_msg = ('=== Added directory %s ===\n' % repo)
            diff_msg = diff_msg + new_msg + '\n'
    else:
        subject = subject + "-"
        for fn, old, new in subjs:
            subject = subject + " %s:%s" % (fn, new)
            if new == 'NONE':
                new_msg = ('=== Removed File %s/%s ===\n'
                           % (repo, fn))
            else:
                try:
                    new_msg = "\n" + create_diff(repo, fn, old, new)
                except IOError:
                    if DRYRUN:
                        text = "[Contrived diff]"
                    else:
                        raise
                    

            diff_msg = diff_msg + new_msg + '\n'

    try:
        # Prepend the Subject and From lines, and append the diff:
        mf = open(msgfilenm, 'r')
        text = mf.read() + diff_msg
        mf.close()
    except IOError:
        if DRYRUN and WAIT:
            text = "[Contrived content]\n" + diff_msg
        else:
            raise

    send_mail(addrs, text, subject)

def create_diff(repo, file, old, new):
    """ Create a diff comparing old and new versions """
    if old == 'NONE':   # A new file was added
        # We have to change to a neutral dir, or cvs will complain about
        # doing checkouts into the repository - even though the checkout
        # is to stdout, sigh.
        origdir = os.getcwd()
        try:
            os.chdir('/tmp')

            revclause = ''
            if new and new != 'NONE':
                revclause = '-r ' + new

            # "Checkout" to stdout, so we can collect the lines to return.
            co_stdout_cmd = 'cvs -fn co -p %s %s/%s' % (revclause, repo, file)
            handle = os.popen(co_stdout_cmd)
            lines = handle.readlines()
            handle.close()
            header = ("=== Added File %s/%s ==="
                      % (repo, file))

        finally:
            os.chdir(origdir)

    else:                # A "normal" update happened
        diff_cmd = ('cvs -d %s -f rdiff -r %s -r %s -kk -u %s/%s'
                    % (CVSROOT_ABS, old, new, repo, file))
        file_handle = os.popen(diff_cmd)
        lines = file_handle.readlines()[2:]
        file_handle.close()
        header = ("=== %s/%s %s => %s ==="
                  % (str(repo), str(file), old, new))

    # Provide for initial checkins on branch - in which case, the ,v files
    # exist only in the Attic.
    if (old == 'NONE') and (len(new.split('.')) > 2):
        template = "%s/%s/Attic/%s,v"
    else:
        template = "%s/%s/%s,v"
    commav = template % (CVSROOT_ABS, repo, file)
    if os.path.exists(commav):
        isbinary = expands_as_binary(commav)
    else:
        complain("Couldn't check binary-ness of missing comma-v"
                 " for %s%s v %s, should be at:\n  %s"
                 % (((old == 'NONE') and "new ") or '',
                    file, new, commav))
        # Let the diff go ahead:
        isbinary = None
    
    if isbinary:
        return "%s\n  <Binary-ish file>" % header
    else:
        total = len(lines)
        if total >= FILE_LINES_LIMIT:
            omitted = total - (2 * FILE_EXCERPTS_LINES)
            # Limit exceeded, show only exercpts from beginning and end.
            lines = (lines[:FILE_EXCERPTS_LINES]
                     + ['\n',
                        ('[-=- -=- -=- %s lines omitted -=- -=- -=-]\n'
                         % (total - (2 * FILE_EXCERPTS_LINES))),
                        '\n']
                     + lines[-FILE_EXCERPTS_LINES:])
            header = header + " (%s/%s lines abridged)" % (omitted, total)
        lines.insert(0, header + "\n")
        return string.join(lines, '')

def do_special(trigger, action, addrs):
    """Do special action - a script, to be invoked from the CVSROOT dir.

    Run it with the version info and send the result to indicated addrs."""

    action_cmd = "%s/CVSROOT/%s %s" % (CVSROOT, action, trigger)
    file_handle = os.popen(action_cmd)

    output = file_handle.read()
    result = file_handle.close()

    if result:
        note_failure("*** Special command failed with error %s:\n"
                     "** %s **\n%s\n",
                     result, action_cmd, output)
    else:
        subject = "CVS: %s run for %s changes" % (action, trigger)
        send_mail(addrs, output, subject)
        complain("%s\n%s\n", subject, output)

def send_mail(addrs, text, subject):
    user = getuser()
    fullname = email = ''
    if user:
        fullname, email = get_user_ldap_info(user)
        if '.' in fullname or ',' in fullname:
            fullname = '"' + fullname + '"'
    else:
        user = "*unknown*"
    if not fullname:
        fullname = "The Unidentified User"
    if not email:
        email = OFFICIAL_SENDER

    cmd_info = {'verbose1': (VERBOSE and "set -x; ") or "",
                'dryrun': (DRYRUN and "echo Would do: ") or "",
                'mailcmd': MAIL_CMD % email,
                'verbose2': (VERBOSE and "-v") or ""}
    cmd = ("%(verbose1)s%(dryrun)s%(mailcmd)s %(verbose2)s" % cmd_info)

    if VERBOSE:
        complain("%sDoing mail cmd for user %s:\n\t%s\n",
                 ((DRYRUN and "NOT ") or ""), user, cmd)

    envelope_info = {'subject': subject,
                     'to': string.join(addrs, ", "),
                     'from': "%s <%s>" % (fullname, email),
                     'sender': OFFICIAL_SENDER,
                     'user': user}
    header = ("Subject: %(subject)s\n"
              "To: %(to)s\n"
              "From: %(from)s\n"
              "Sender: %(sender)s\n"
              % envelope_info)

    notice = header + '\n' + text

    if not DRYRUN:
        cmd_in, cmd_out = os.popen2(cmd, 'rw')
        cmd_in.write(notice)
        cmd_in.close()
        output = cmd_out.read()
        result = cmd_out.close()
    else:
        result = None

    if VERBOSE:
        complain(string.join(map(lambda x: '= ' + x,
                                 string.split(header, '\n')),
                             '\n'))
    if result:
        note_failure("*** Mail cmd yielded unexpected result %s:\n%s\n",
                     result, output)

def loosen_file(fname):
    """Relax permissions on (newly created) file so others can use it too."""
    try:
        os.chmod(fname, 0777)
    except os.error, err:
        pass

def grok_file(s):
    """Separate "file,old-version,new-version"."""
    m = sVv_re.match(s)
    if not m:
        raise ValueError, "'%s' not in file,old-vers,new-vers format" % s
    return m.groups()

failures = 0
def note_failure(msg, *args):
    """Register a failure for handle_failures to announce at the end."""
    global failures
    failures = 1
    apply(complain, (msg,) + args)

def getuser():
    """Try to get the user's login name."""
    try: return getpass.getuser()
    except: return None

def get_user_ldap_info(user):
    """Obtain some aproximation to user's (fullname, email).

    We prefer to get the info out of ldap.

    Failing that, we use whatever we can get from the password file.

    Failing that, we return an empty fullname and the value of OFFICIAL_SENDER.

    Failling all, we return ('', '')."""
    if not user:
        return ('', '')
    # Fallback values:
    # 003
    #email = OFFICIAL_SENDER
    email = "%s@%s" % (user, MY_DOMAIN)

    try:
        fullname = pwd.getpwnam(user)[4]
    except KeyError:
        fullname = ''

    if USE_LDAP:
        try:
            import ldap; ldap_mod = ldap
        except ImportError:
            try:
                import _ldap; ldap_mod = _ldap
            except ImportError:
                print "Failed to get any LDAP module, punting on ldap."
                return (fullname, email)
        try:
            c = ldap_mod.open(LDAP_HOST, LDAP_PORT)
            c.simple_bind_s('', '')
            # ('sn' for "surname")
            record = c.search_s('ou=people,o=DC,c=US',
                                ldap_mod.SCOPE_SUBTREE,
                                'loginName=%s' % user,
                                ['mail', 'sn', 'givenName'])
            if record:
                d = record[0][1]
                email = d.get('mail', [email])[0]
                first = d.get('givenName', [''])[0]
                last = d.get('sn', [''])[0]
                if first or last:
                    if first: first = first + " "
                    fullname = "%s%s" % (first, last)
        except ldap_mod.LDAPError:
            pass

    return (fullname, email)

def handle_failures(argstring):
    """On serious failures, send the complaints log to CVSMASTER."""

    if os.environ.has_key('HOSTNAME'):
        host = os.environ['HOSTNAME']
    else:
        host = socket.gethostbyaddr(socket.gethostname())[0]
        if not host:
            host = "nohost.zope.com"

    user = getuser() or "unidentified"

    if os.path.isfile(OUTPUT_LOG):
        log_file_expr = ("\n\tSee log file for application errors:\n\t\t%s"
                         % OUTPUT_LOG)
    else:
        log_file_expr = ""

    complain("Sending complaints log to CVSMASTER %s\n", CVSMASTER)
    complain("Time stamp: %s\n", time.ctime(time.time()))
    complain("Fatal failures in %s:%s"
             "\n\tCheckin by: %s@%s\n\tInvocation: %s\n\n",
             SCRIPT, log_file_expr, user, host, argstring)

    cmd = ('%s %s -s "CVS errors in %s for %s" %s'
           % (MAIL_CMD, ((VERBOSE and "-v") or ""), SCRIPT, user, CVSMASTER))
    f = os.popen(cmd, 'w')
    f.write("Serious errors encountered during CVS postcommit actions,\n")
    f.write("the log is below with orienting details at the bottom.\n\n")
    f.write(string.join(complaints, ""))
    return f.close()

def usage():
    complain("Usage: %s [options] ", SCRIPT)
    complain('"repopath file,oldv,newv [f2,o2,n2 ...]"\n')
    complain("(Note that repo-path and files must be"
             " quoted so they are a single token.)\n")

def get_command(cmd, path):
    """Get a valid exe for cmd on path."""
    for d in path:
        maybe = os.path.join(d, cmd)
        if os.path.isfile(maybe):
            return maybe
    note_failure("No usable %s executable found.\n", `cmd`)
    
EXPANDS_RE = re.compile("expand.*@b@")
def expands_as_binary(repository_path):
    """Return true if the repository file is marked for binary keyword
    expansion."""

    f = open(repository_path, 'r')
    while 1:
        l = f.readline().strip()
        if not l:
            # Break out on first blank line or eof:
            return 0
        if EXPANDS_RE.match(l):
            return 1

complaints = []
def complain(msg, *args):
    global complaints

    import time
    t = time.strftime("%Y-%m-%d %T - ", time.localtime(time.time()))

    complaints.append(t + (msg % args))
    sys.stderr.write(msg % args)
    sys.stderr.flush()

def safe_getcwd(fallback):
    try:
        return os.getcwd()
    except:
        os.chdir(fallback)
        try:
            return os.getcwd()
        except:
            return "<unknown dir>"

if __name__ == "__main__":
    main(sys.argv)