#! /usr/bin/env python
#
# $YahooCvsId: bug2dot.py,v 1.6 2008/11/24 17:02:56 Exp $
# $Source: /CVSROOT/.../bug2dot/src/bug2dot.py,v $
#
# Copyright (c) 2008 Yahoo! Inc.
#
# Originally written by Jan Schaumann <jschauma@> in August 2008.
#
# This program emits a dot language description of the dependency tree for
# the given bugs.  It is very crudely hacked together, as it's supposed to
# be a temporary solution until 
# http://<bugzilla-url>/show_bug.cgi?id=2003070 is addressed.
#
# Yes, I know that there's nothing as permanent as a temporary solution.
#
# Please file bugs and feature requests at:
# http://<bugzilla-url>/enter_bug.cgi?product=<product>&component=bug2dot

import getopt
from os.path import basename
import re
import string 
import sys
import time

import urllib2

###
### Globals
###

EXIT_ERROR = 1
EXIT_SUCCESS = 0

# "private"
__BASEURL = 'http://<bugzilla-url>/showdependencytree.cgi?id='
__NODE2COLOR = { 'resolved':'green', 'top-level':'black', 'open':'red' }
__NUMBERS_ONLY = False
__RESOLVED_BUGS = {}
__OPEN_ONLY = False
__VERSION = "0.1"

###
### Subroutines
###

def dehtmlify(input):
    """crudely remove html tags"""

    output = re.sub('<.*?>', '', input.strip())
    output = re.sub('&quot;', '\\\"', output)

    return output


def dotFooter():
    """print the end of the dot file"""

    print '}'
    print '// %s' % time.ctime()
    print '// end dot description generated by bug2dot version %s' % __VERSION


def dotHeader():
    """print the beginning for the dot file"""

    print '// begin dot description generated by bug2dot version %s' % __VERSION
    print '// %s' % " ".join(sys.argv)
    print '// %s' % time.ctime()
    print 'digraph "bugs" {'


def dotify(bug):
    """fetch the dependency tree for the given bug and generate a dot
    language description on stdout"""

    global __NUMBERS_ONLY, __RESOLVED_BUGS

    current = bug
    inLink = 0
    nextSummary = False
    resolved = False
    reverse = False
    parent = bug
    grandparent = bug

    bugs = [ bug ]
    edgesSeen = []
    nodesSeen = []

    print '// --- Begin %s ---' % bug

    if __NUMBERS_ONLY:
        printNode(bug, 'top-level', 'Bug ' + bug)
        nodesSeen.append(bug)

    url = urllib2.urlopen(__BASEURL + bug)

    for line in url:

        if not __NUMBERS_ONLY:
            if (line.find('<h1>Dependency tree for ') != -1):
                nextSummary = True
                continue

            if nextSummary:
                desc = "Bug " + bug + ": " + dehtmlify(line)
                printNode(bug, 'top-level', desc)
                nodesSeen.append(bug)
                nextSummary = False
       
        if (line.find('blocks</h3>') != -1):
            reverse = True

        if (line.find('<ul>') != -1):
            grandparent = parent
            parent = current

        if (line.find('</ul>') != -1):
            parent = grandparent
            grandparent = bug
            if not bugs:
                parent = bug
                break

        if (line.find('<span class="bz_closed">') != -1):
            resolved = True

        if (line.find('<a href="show_bug.cgi?id=') != -1):
            bp = re.compile('.*<a href="show_bug.cgi\?id=(?P<child>[0-9]+)">[0-9]+')
            rem = bp.match(line)

            if (rem):
                current = rem.group('child')
                bugs.append(current)
                inLink = 1
                continue

        if (inLink == 1):
            inLink += 1
            continue

        if (line.find('</li>') != -1):
            if len(bugs) > 1:
                previous = bugs.pop()
                last = bugs[len(bugs) - 1]
            else:
                continue

            if reverse:
                printEdge(bug, last, previous, edgesSeen)
            else:
                printEdge(bug, previous, last, edgesSeen)


        if (inLink == 2):
            type = "open"
            if (resolved):
                type = "resolved"
                __RESOLVED_BUGS[current] = 1

            if not current in nodesSeen:
                nodesSeen.append(current)
                desc = "Bug " + current
                if not __NUMBERS_ONLY:
                    desc = desc + ": " + dehtmlify(line)

                printNode(current, type, desc)

            if current != parent:
                if reverse:
                    if len(bugs) != 2:
                        printEdge(bug, parent, current, edgesSeen)
                else:
                    printEdge(bug, current, parent, edgesSeen)

            resolved = False
            inLink = 0

    url.close()
    print '// --- End %s ---' % bug


def handleStdin():
    """read items from stdin, split by whitespace, then traverse"""

    rl = sys.stdin.readlines
    lines = rl(1024)
    while lines:
        for line in lines:
            for word in line.split():
                dotify(word)

        lines = rl(1024)


def main():
    """parse command-line options and act on them"""

    args = parseOptions()

    if not args:
        usage()
        sys.exit(EXIT_ERROR)
        # NOTREACHED

    dotHeader()

    for a in args:
        if a == '-':
            handleStdin()
        else:
            dotify(a)

    dotFooter()

    return EXIT_SUCCESS


def parseOptions():
    """parse command-line options via getopt, return remaining args"""

    global __NUMBERS_ONLY, __OPEN_ONLY

    try:
        opts, args = getopt.getopt(sys.argv[1:], "hno")
    except getopt.GetoptError:
        usage()
        sys.exit(EXIT_ERROR)
        # NOTREACHED

    for o, a in opts:
        if o in ("-h"):
            usage()
            sys.exit(EXIT_SUCCESS)
            # NOTREACHED
        if o in ("-n"):
            __NUMBERS_ONLY = True
        if o in ("-o"):
            __OPEN_ONLY = True

    return args



def printEdge(top, a, b, seen):
    """print an edge entry from a to b iff not in seen; append tuple to seen"""

    global __OPEN_ONLY, __RESOLVED_BUGS

    style = ''

    if __OPEN_ONLY:
        if a in __RESOLVED_BUGS:
            return
        if b in __RESOLVED_BUGS:
            style = ',style="dotted"'
            b = top

    t = (a, b)
    if not t in seen:
        seen.append(t)
        print '"%s" -> "%s" [color="black"%s];  // EDGE' % (a, b, style)


def printNode(a, type, desc):
    """print a node entry"""

    global __NODE2COLOR, __OPEN_ONLY

    if (type == "resolved") and __OPEN_ONLY:
        return

    print '"%s" [color="%s",label="%s"]; // NODE' % (a, __NODE2COLOR[type], desc)


def usage():
    """print short usage"""

    print 'Usage: %s [-hno] bug [bug [...]]' % basename(sys.argv[0])
    print '\t-h    Print a usage statement and exit.'
    print '\t-n    Only use numbers for all bugs.'
    print '\t-o    Only display open bugs.'


###
### Main
###

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