#! /usr/bin/env python

import os, sys, logging, re, collections

##@var USHhwrf
# The ush/ subdirectory of the HWRF installation directory
USHhwrf=None

##@var HOMEhwrf
# The HWRF installation directory
HOMEhwrf=None

if os.environ.get('USHhwrf',''):   USHhwrf=os.environ['USHhwrf']
if os.environ.get('HOMEhwrf',''):  HOMEhwrf=os.environ['HOMEhwrf']

if HOMEhwrf is None and USHhwrf is None:
    HOMEhwrf=os.path.dirname(os.getcwd())
    USHguess=os.path.join(HOMEhwrf,'ush')
    if os.path.isdir(USHguess):
        if USHhwrf is None: USHhwrf=USHguess

if HOMEhwrf is not None:
    if USHhwrf is None:            USHhwrf=os.path.join(HOMEhwrf,'ush')

if USHhwrf is None: 
    print>>sys.stderr, "Cannot guess $USHhwrf.  Please set $HOMEhwrf or " \
        "$USHhwrf in environment."
    sys.exit(2)

sys.path.append(USHhwrf)

import hwrf.numerics
from hwrf.numerics import to_datetime,to_datetime_rel, to_fraction,to_timedelta
import produtil.run, produtil.setup
from produtil.run import runstr, batchexe

produtil.setup.setup(send_dbn=False)

epsilon=to_timedelta(30) # 30 seconds
six_hours=to_timedelta(3600*6)
nfile=0
for xmlfile in sys.argv[1:]:
    nfile+=1
    #print '%s: SCAN FILE'%(xmlfile,)
    cycles=set()
    is_xml=False
    with open(xmlfile,'rt') as xmlf:
        for line in xmlf:
            if re.search('<?xml version=',line):
                is_xml=True
            m=re.search('<cycledef>([0-9]+)</cycledef>',line)
            if m and m.groups(1) and m.groups(1)[0]:
                #print line
                cycle=to_datetime(m.groups(1)[0])
                cycles.add(cycle.strftime('%Y%m%d%H%M'))
            else:    
                m=re.search('<cycledef>([0-9]+) ([0-9]+) 0?6:00:00</cycledef>',line)
                if m and m.groups(1) and m.groups(1)[0] and m.groups(1)[1]:
                    #print line
                    start=to_datetime(m.groups(1)[0])
                    end=to_datetime(m.groups(1)[1])
                    now=start
                    stop=end+epsilon
                    while now<stop:
                        cycles.add(now.strftime('%Y%m%d%H%M')) 
                        now=to_datetime_rel(six_hours,now)
                elif re.search('<cycledef>',line):
                    print 'BAD CYCLEDEF: '+line
                    exit(1)
    if not is_xml:
        print '%s: NOT AN XML FILE!!'%(xmlfile,)
        exit(1)
    #print 'Cycles: '+( ' '.join(sorted(cycles)))
    dbfile=xmlfile[:-3]+'db'
    cmd=batchexe('rocotostat')['-w',xmlfile,'-d',dbfile,'-c','ALL']
    #print repr(cmd)
    stat=runstr(cmd)
    last=dict()
    complete=set()
    queued=set()
    badjobs=set()
    badincompletejobs=set()
    running=set()
    printme=list()
    failme=collections.defaultdict(dict)
    for line in stat.splitlines():
        #201408061200  completion  315615  SUCCEEDED
        m=re.match('^\s*([0-9]{12})\s+(\S+)\s+\S+\s+(\S+)',line)
        if m and m.groups(1) and m.groups(1)[0] and m.groups(1)[1] and m.groups(1)[1] and m.groups(1)[2]:
            cycle=m.groups(1)[0]
            job=m.groups(1)[1]
            status=m.groups(1)[2]
            if status=='-': continue # ignore unstarted jobs
            if status=='SUCCEEDED':
                if job=='completion':
                    complete.add(m.groups(1)[0])
                last[cycle]=job
            elif status in ('DEAD','UNKNOWN','LOST','UNAVAILABLE'):
                failme[cycle][job]=status
                badjobs.add(cycle)
            elif status=='RUNNING':
                running.add(cycle)
            else:
                queued.add(cycle)
    lastline=''
    lastcycle=''
    prevcycle=''
    for cycle in sorted(cycles):
        if cycle not in last:
            line='unstarted.'
        elif cycle in complete:
            line='complete.'
        else:
            line='[last job finished: '+last[cycle]+']'
        if cycle in running: line+=' [RUNNING]'
        if cycle in queued: line+=' [QUEUED]'
        if cycle in badjobs: line+=' [FAILURES]'
        #printme.append('%s: %s(%s-%s): %s'%(xmlfile,cycle,lastcycle,prevcycle,line))
        if lastline and lastline != line:
            if prevcycle and prevcycle!=lastcycle:
                printme.append('%s-%s: %s'%(lastcycle,prevcycle,lastline))
            else:
                printme.append('             %s: %s'%(lastcycle,lastline))
            lastline=line
            lastcycle=cycle
            prevcycle=cycle
        elif lastline and lastline==line:
            #printme.append('lastline and lastline!=line')
            prevcycle=cycle
            lastline=line
        else:
            #printme.append('ELSE')
            prevcycle=cycle
            lastcycle=cycle
            lastline=line
    if prevcycle and prevcycle!=lastcycle:
        printme.append('%s-%s: %s'%(lastcycle,prevcycle,lastline))
    elif lastcycle:
        printme.append('             %s: %s'%(lastcycle,lastline))

    incomplete=cycles-complete

    sent_heading=False
    for cycle in sorted(cycles):
        appendme=''
        is_incomplete=cycle in complete
        if is_incomplete:
            appendme=' (but cycle completed)'
        for (job,status) in failme[cycle].iteritems():
            if not is_incomplete:
                badincompletejobs.add(cycle)
            if not sent_heading:
                printme.append('Failed jobs:')
                sent_heading=True
            printme.append('%s %s IS %s%s'%(
                cycle,job,status,appendme))

    if nfile>1:
        print

    if not running and not queued and not badjobs and incomplete:
        print xmlfile+': STALLED.'
        print 'Nothing queued, failed or running, but workflow is incomplete.  You'
        print 'should check for dependency errors, and ensure your cron job is running.'
        for line in printme: print line
    elif not incomplete:
        print xmlfile+': COMPLETE.'
    elif badincompletejobs:
        print xmlfile+': FAILED JOBS!'
        for line in printme: print line
    elif running: 
        if badjobs:
            print xmlfile+': RUNNING (failed jobs for completed cycles).'
        else:
            print xmlfile+': RUNNING.'
        for line in printme: print line
    elif queued:
        if badjobs:
            print xmlfile+': QUEUED (failed jobs for completed cycles).'
        else:
            print xmlfile+': QUEUED.'
        for line in printme: print line
    else:
        print xmlfile+':'
        for line in printme: print line

#     incomplete=cycles-complete
#     if incomplete:
#         print xmlfile+': incomplete cycles: '+(' '.join(sorted(incomplete)))
#     else:
#         print xmlfile+': all cycles completed.'