CRAB/python/crab_util.py

###########################################################################
#
#   C O N V E N I E N C E    F U N C T I O N S
#
###########################################################################

import string, sys, os, time
import ConfigParser, re, popen2, select, fcntl

import common
from crab_exceptions import CrabException

###########################################################################
def parseOptions(argv):
    """
    Parses command-line options.
    Returns a dictionary with specified options as keys:
    -opt1             --> 'opt1' : None
    -opt2 val         --> 'opt2' : 'val'
    -opt3=val         --> 'opt3' : 'val'
    Usually called as
    options = parseOptions(sys.argv[1:])
    """
    options = {}
    argc = len(argv)
    i = 0
    while ( i < argc ):
        if argv[i][0] != '-':
            i = i + 1
            continue
        eq = string.find(argv[i], '=')
        if eq > 0 :
            opt = argv[i][:eq]
            val = argv[i][eq+1:]
            pass
        else:
            opt = argv[i]
            val = None
            if ( i+1 < argc and argv[i+1][0] != '-' ):
                i = i + 1
                val = argv[i]
                pass
            pass
        options[opt] = val
        i = i + 1
        pass
    return options

def loadConfig(file):
    """
    returns a dictionary with keys of the form
    <section>.<option> and the corresponding values
    """
    config={}
    cp = ConfigParser.ConfigParser()
    cp.read(file)
    for sec in cp.sections():
        # print 'Section',sec
        for opt in cp.options(sec):
            #print 'config['+sec+'.'+opt+'] = '+string.strip(cp.get(sec,opt))
            config[sec+'.'+opt] = string.strip(cp.get(sec,opt))
    return config

###########################################################################
def isInt(str):
    """ Is the given string an integer ?"""
    try: int(str)
    except ValueError: return 0
    return 1

###########################################################################
def isBool(str):
    """ Is the given string 0 or 1 ?"""
    if (str in ('0','1')): return 1
    return 0

###########################################################################
def parseRange(range):
    """
    Takes as the input a string with two integers separated by
    the minus sign and returns the tuple with these numbers:
    'n1-n2' -> (n1, n2)
    'n1'    -> (n1, n1)
    """
    start = None
    end   = None
    minus = string.find(range, '-')
    if ( minus < 0 ):
        if isInt(range):
            start = int(range)
            end   = start
            pass
        pass
    else:
        if isInt(range[:minus]) and isInt(range[minus+1:]):
            start = int(range[:minus])
            end   = int(range[minus+1:])
            pass
        pass
    return (start, end)

###########################################################################
def parseRange2(range):
    """
    Takes as the input a string in the form of a comma-separated
    numbers and ranges
    and returns a list with all specified numbers:
    'n1'          -> [n1]
    'n1-n2'       -> [n1, n1+1, ..., n2]
    'n1,n2-n3,n4' -> [n1, n2, n2+1, ..., n3, n4]
    """
    result = []
    if not range: return result

    comma = string.find(range, ',')
    if comma == -1: left = range
    else:           left = range[:comma]

    (n1, n2) = parseRange(left)
    while ( n1 <= n2 ):
        try:
            result.append(n1)
            n1 += 1
            pass
        except:
            msg = 'Syntax error in range <'+range+'>'
            raise CrabException(msg)

    if comma != -1:
        try:
            result.extend(parseRange2(range[comma+1:]))
            pass
        except:
            msg = 'Syntax error in range <'+range+'>'
            raise CrabException(msg)

    return result

###########################################################################
def crabJobStatusToString(crab_status):
    """
    Convert one-letter crab job status into more readable string.
    """
    status={
           'C':'Created',
           'D':'Done',
           'H':'Hold',
           'U':'Ready',
           'I':'Scheduled',
           'X':'Canceled',
           'W':'Created',
           'R':'Running',
           'SC':'Checkpointed',
           'SS':'Scheduled',
           'SR':'Ready',
           'RE':'Ready',
           'SW':'Waiting',
           'SU':'Submitted',
           'S' :'Submitted (Boss)',
           'UN':'Undefined',
           'SK':'Cancelled',
           'SD':'Done (Success)',
           'SA':'Aborted',
           'DA':'Done (Aborted)',
           'SE':'Cleared',
           'OR':'Done (Success)',
           'A?':'Aborted',
           'K':'Killed',
           'E':'Cleared',
           'Z':'Cleared (Corrupt)',
           'NA':'Unknown',
           'I?':'Idle',
           'O?':'Done',
           'R?':'Running'
           }
#    if   crab_status == 'C': status = 'Created'
#    elif crab_status == 'D': status = 'Done'
#    elif crab_status == 'R': status = 'Submitted'#Should be running? ds  
#    elif crab_status == 'S': status = 'Submitted'
#    elif crab_status == 'K': status = 'Killed'
#    elif crab_status == 'X': status = 'None'
#    elif crab_status == 'Y': status = 'Output retrieved'
#    elif crab_status == 'A': status = 'Aborted'
#    elif crab_status == 'RC': status = 'ReCreated'
#    else: status = '???'
    return status[crab_status]

###########################################################################
def findLastWorkDir(dir_prefix, where = None):

    if not where: where = os.getcwd() + '/'
    # dir_prefix usually has the form 'crab_0_'
    pattern = re.compile(dir_prefix)

    file_list = []
    for fl in os.listdir(where):
        if pattern.match(fl):
            file_list.append(fl)
            pass
        pass

    if len(file_list) == 0: return None

    file_list.sort()

    wdir = where + file_list[len(file_list)-1]
    return wdir

###########################################################################
def importName(module_name, name):
    """
    Import a named object from a Python module,
    i.e., it is an equivalent of 'from module_name import name'.
    """
    module = __import__(module_name, globals(), locals(), [name])
    return vars(module)[name]


###########################################################################

### WARNING This Function become USELESS after Boss API implementation
def runBossCommand(cmd, printout=0, timeout=3600):
    """
    Cd to correct directory before running a boss command
    """
    cwd = os.getcwd()
    os.chdir(common.work_space.shareDir())
    out = runCommand(cmd, printout, timeout)
    os.chdir(cwd)
    return out

###########################################################################
def readable(fd):
    return bool(select.select([fd], [], [], 0)) 

###########################################################################
def makeNonBlocking(fd):
    fl = fcntl.fcntl(fd, fcntl.F_GETFL)
    try:
        fcntl.fcntl(fd, fcntl.F_SETFL, fl | os.O_NDELAY)
    except AttributeError:
            fcntl.fcntl(fd, fcntl.F_SETFL, fl | os.FNDELAY)

###########################################################################
def runCommand(cmd, printout=0, timeout=-1):
    """
    Run command 'cmd'.
    Returns command stdoutput+stderror string on success,
    or None if an error occurred.
    Following recipe on http://aspn.activestate.com/ASPN/Cookbook/Python/Recipe/52296
    """

    if printout:
        common.logger.message(cmd)
    else:
        common.logger.debug(10,cmd)
        common.logger.write(cmd)
        pass

    child = popen2.Popen3(cmd, 1) # capture stdout and stderr from command
    child.tochild.close()             # don't need to talk to child
    outfile = child.fromchild 
    outfd = outfile.fileno()
    errfile = child.childerr
    errfd = errfile.fileno()
    makeNonBlocking(outfd)            # don't deadlock!
    makeNonBlocking(errfd)
    outdata = []
    errdata = []
    outeof = erreof = 0

    if timeout > 0 :
        maxwaittime = time.time() + timeout

    err = -1
    while (timeout == -1 or time.time() < maxwaittime):
        ready = select.select([outfd,errfd],[],[]) # wait for input
        if outfd in ready[0]:
            outchunk = outfile.read()
            if outchunk == '': outeof = 1
            outdata.append(outchunk)
        if errfd in ready[0]:
            errchunk = errfile.read()
            if errchunk == '': erreof = 1
            errdata.append(errchunk)
        if outeof and erreof:
            err = child.wait()
            break
        select.select([],[],[],.01) # give a little time for buffers to fill
    if err == -1:
        # kill the pid
        common.logger.message('killing process '+(cmd)+' with timeout '+str(timeout))
        os.kill (child.pid, 9)
        err = child.wait()

    cmd_out = string.join(outdata,"")
    cmd_err = string.join(errdata,"")

    if err:
        common.logger.message('`'+cmd+'`\n   failed with exit code '
                           +`err`+'='+`(err&0xff)`+'(signal)+'
                           +`(err>>8)`+'(status)')
        common.logger.message(cmd_out)
        common.logger.message(cmd_err)
        return None

#    cmd_out = string.join(outdata,"")
#    cmd_err = string.join(errdata,"")
    cmd_out = cmd_out + cmd_err
    if printout:
        common.logger.message(cmd_out)
    else:
        common.logger.debug(10,cmd_out)
        common.logger.write(cmd_out)
        pass
    #print "<"+cmd_out+">"
    return cmd_out

####################################
def makeCksum(filename) :
    """
    make chksum using filename and content of file
    """

    tobedeleted=0
    try:
        import tempfile
        tmpfile= tempfile.NamedTemporaryFile(mode='w')
        tmp_filename = tmpfile.name
    except:
        ## SL in case py2.2 is used, fall back to old solution
        tmp_filename = tempfile.mktemp()
        tmpfile= open(tmp_filename,'w')
        tobedeleted=1

    # add filename as first line
    tmpfile.write(filename+'\n')

    # fill input file in tmp file
    infile = open(filename)
    tmpfile.writelines(infile.readlines())
    tmpfile.flush()
    infile.close()

    cmd = 'cksum '+tmp_filename
    cmd_out = runCommand(cmd)

    cksum = cmd_out.split()[0]

    ## SL this is not needed if we use NamedTemporaryFile, which is not available in py2.2
    if (tobedeleted): os.remove(tmp_filename)

    return cksum

def spanRanges(jobArray):
    """
    take array of job numbers and concatenate 1,2,3 to 1-3
    return string
    """

    output = ""
    jobArray.sort()
        
    previous = jobArray[0]-1
    for job in jobArray:
        if previous+1 == job:
            previous = job
            if len(output) > 0 :
                if output[-1] != "-":
                    output += "-"
            else :
                output += str(previous)
        else:
            output += str(previous) + "," + str(job)
            #output += "," + str(job)
            previous = job
    if len(jobArray) > 1 :
        output += str(previous)

    return output

####################################
if __name__ == '__main__':
    print 'sys.argv[1] =',sys.argv[1]
    list = parseRange2(sys.argv[1])
    print list
    cksum = makeCksum("crab_util.py")
    print cksum
    
Revision:	1.38
Committed:	Wed Feb 13 18:17:54 2008 UTC (17 years, 2 months ago) by slacapra
Content type:	text/x-python
Branch:	MAIN
CVS Tags:	CRAB_2_1_2, CRAB_2_1_2_pre2, CRAB_2_1_2_pre1, CRAB_2_1_1, CRAB_2_1_1_pre3, CRAB_2_1_1_pre1, CRAB_2_1_0, CRAB_2_1_0_pre6, CRAB_2_1_0_pre5
Branch point for:	CRAB_2_1_2_br, CRAB_2_1_1_pre2
Changes since 1.37:	+2 -0 lines
Log Message:	Fix problem in resubmission when passing a parameter the number of the job you want to resubmit eg -resubmit 4 now means (as was before) to resubmit job #4, not to resubmit 4 jobs. Create a new class Resubmitter (derived from Submitter) to deal with the logic of selecting the job to be resubmitetd from the user request and job status. This clean up a bit crab.py The actual submission is always done by Submitter::run()
#	User	Rev	Content
1	nsmirnov	1.1	###########################################################################
2			#
3			# C O N V E N I E N C E F U N C T I O N S
4			#
5			###########################################################################
6
7	slacapra	1.12	import string, sys, os, time
8	slacapra	1.16	import ConfigParser, re, popen2, select, fcntl
9	nsmirnov	1.1
10	nsmirnov	1.2	import common
11	slacapra	1.30	from crab_exceptions import CrabException
12	nsmirnov	1.1
13			###########################################################################
14			def parseOptions(argv):
15			"""
16			Parses command-line options.
17			Returns a dictionary with specified options as keys:
18			-opt1 --> 'opt1' : None
19			-opt2 val --> 'opt2' : 'val'
20			-opt3=val --> 'opt3' : 'val'
21			Usually called as
22			options = parseOptions(sys.argv[1:])
23			"""
24			options = {}
25			argc = len(argv)
26			i = 0
27			while ( i < argc ):
28			if argv[i][0] != '-':
29			i = i + 1
30			continue
31			eq = string.find(argv[i], '=')
32			if eq > 0 :
33			opt = argv[i][:eq]
34			val = argv[i][eq+1:]
35			pass
36			else:
37			opt = argv[i]
38			val = None
39			if ( i+1 < argc and argv[i+1][0] != '-' ):
40			i = i + 1
41			val = argv[i]
42			pass
43			pass
44			options[opt] = val
45			i = i + 1
46			pass
47			return options
48
49			def loadConfig(file):
50			"""
51			returns a dictionary with keys of the form
52			<section>.<option> and the corresponding values
53			"""
54			config={}
55			cp = ConfigParser.ConfigParser()
56			cp.read(file)
57			for sec in cp.sections():
58			# print 'Section',sec
59			for opt in cp.options(sec):
60			#print 'config['+sec+'.'+opt+'] = '+string.strip(cp.get(sec,opt))
61			config[sec+'.'+opt] = string.strip(cp.get(sec,opt))
62			return config
63
64			###########################################################################
65			def isInt(str):
66			""" Is the given string an integer ?"""
67			try: int(str)
68			except ValueError: return 0
69			return 1
70
71			###########################################################################
72			def isBool(str):
73			""" Is the given string 0 or 1 ?"""
74			if (str in ('0','1')): return 1
75			return 0
76
77			###########################################################################
78	nsmirnov	1.3	def parseRange(range):
79			"""
80			Takes as the input a string with two integers separated by
81			the minus sign and returns the tuple with these numbers:
82			'n1-n2' -> (n1, n2)
83			'n1' -> (n1, n1)
84			"""
85			start = None
86			end = None
87			minus = string.find(range, '-')
88			if ( minus < 0 ):
89			if isInt(range):
90			start = int(range)
91			end = start
92			pass
93			pass
94			else:
95			if isInt(range[:minus]) and isInt(range[minus+1:]):
96			start = int(range[:minus])
97			end = int(range[minus+1:])
98			pass
99			pass
100			return (start, end)
101
102			###########################################################################
103	nsmirnov	1.4	def parseRange2(range):
104			"""
105			Takes as the input a string in the form of a comma-separated
106			numbers and ranges
107			and returns a list with all specified numbers:
108			'n1' -> [n1]
109			'n1-n2' -> [n1, n1+1, ..., n2]
110			'n1,n2-n3,n4' -> [n1, n2, n2+1, ..., n3, n4]
111			"""
112	slacapra	1.30	result = []
113			if not range: return result
114	nsmirnov	1.4
115			comma = string.find(range, ',')
116			if comma == -1: left = range
117			else: left = range[:comma]
118
119			(n1, n2) = parseRange(left)
120			while ( n1 <= n2 ):
121	slacapra	1.11	try:
122	slacapra	1.30	result.append(n1)
123	slacapra	1.11	n1 += 1
124			pass
125			except:
126			msg = 'Syntax error in range <'+range+'>'
127			raise CrabException(msg)
128	nsmirnov	1.4
129			if comma != -1:
130	slacapra	1.11	try:
131	slacapra	1.30	result.extend(parseRange2(range[comma+1:]))
132	slacapra	1.11	pass
133			except:
134			msg = 'Syntax error in range <'+range+'>'
135			raise CrabException(msg)
136	nsmirnov	1.4
137	slacapra	1.30	return result
138	nsmirnov	1.4
139			###########################################################################
140	nsmirnov	1.3	def crabJobStatusToString(crab_status):
141			"""
142			Convert one-letter crab job status into more readable string.
143			"""
144	corvo	1.27	status={
145	slacapra	1.38	'C':'Created',
146			'D':'Done',
147	corvo	1.27	'H':'Hold',
148			'U':'Ready',
149			'I':'Scheduled',
150			'X':'Canceled',
151			'W':'Created',
152			'R':'Running',
153			'SC':'Checkpointed',
154			'SS':'Scheduled',
155			'SR':'Ready',
156			'RE':'Ready',
157			'SW':'Waiting',
158			'SU':'Submitted',
159			'S' :'Submitted (Boss)',
160			'UN':'Undefined',
161			'SK':'Cancelled',
162			'SD':'Done (Success)',
163			'SA':'Aborted',
164			'DA':'Done (Aborted)',
165			'SE':'Cleared',
166			'OR':'Done (Success)',
167			'A?':'Aborted',
168			'K':'Killed',
169			'E':'Cleared',
170			'Z':'Cleared (Corrupt)',
171			'NA':'Unknown',
172			'I?':'Idle',
173			'O?':'Done',
174			'R?':'Running'
175			}
176			# if crab_status == 'C': status = 'Created'
177			# elif crab_status == 'D': status = 'Done'
178			# elif crab_status == 'R': status = 'Submitted'#Should be running? ds
179			# elif crab_status == 'S': status = 'Submitted'
180			# elif crab_status == 'K': status = 'Killed'
181			# elif crab_status == 'X': status = 'None'
182			# elif crab_status == 'Y': status = 'Output retrieved'
183			# elif crab_status == 'A': status = 'Aborted'
184			# elif crab_status == 'RC': status = 'ReCreated'
185			# else: status = '???'
186			return status[crab_status]
187	nsmirnov	1.3
188			###########################################################################
189	nsmirnov	1.1	def findLastWorkDir(dir_prefix, where = None):
190
191			if not where: where = os.getcwd() + '/'
192			# dir_prefix usually has the form 'crab_0_'
193			pattern = re.compile(dir_prefix)
194
195			file_list = []
196			for fl in os.listdir(where):
197			if pattern.match(fl):
198			file_list.append(fl)
199			pass
200			pass
201
202			if len(file_list) == 0: return None
203
204			file_list.sort()
205
206			wdir = where + file_list[len(file_list)-1]
207			return wdir
208
209			###########################################################################
210			def importName(module_name, name):
211			"""
212			Import a named object from a Python module,
213			i.e., it is an equivalent of 'from module_name import name'.
214			"""
215			module = __import__(module_name, globals(), locals(), [name])
216			return vars(module)[name]
217
218	spiga	1.9
219	slacapra	1.11	###########################################################################
220	spiga	1.28
221	slacapra	1.29	### WARNING This Function become USELESS after Boss API implementation
222	gutsche	1.24	def runBossCommand(cmd, printout=0, timeout=3600):
223	slacapra	1.11	"""
224			Cd to correct directory before running a boss command
225			"""
226			cwd = os.getcwd()
227			os.chdir(common.work_space.shareDir())
228			out = runCommand(cmd, printout, timeout)
229			os.chdir(cwd)
230			return out
231	spiga	1.9
232	nsmirnov	1.1	###########################################################################
233	slacapra	1.16	def readable(fd):
234			return bool(select.select([fd], [], [], 0))
235
236			###########################################################################
237			def makeNonBlocking(fd):
238			fl = fcntl.fcntl(fd, fcntl.F_GETFL)
239			try:
240			fcntl.fcntl(fd, fcntl.F_SETFL, fl \| os.O_NDELAY)
241			except AttributeError:
242			fcntl.fcntl(fd, fcntl.F_SETFL, fl \| os.FNDELAY)
243
244			###########################################################################
245	slacapra	1.11	def runCommand(cmd, printout=0, timeout=-1):
246	nsmirnov	1.1	"""
247			Run command 'cmd'.
248			Returns command stdoutput+stderror string on success,
249			or None if an error occurred.
250	slacapra	1.16	Following recipe on http://aspn.activestate.com/ASPN/Cookbook/Python/Recipe/52296
251	nsmirnov	1.1	"""
252	fanzago	1.15
253	slacapra	1.11	if printout:
254			common.logger.message(cmd)
255			else:
256	slacapra	1.13	common.logger.debug(10,cmd)
257	slacapra	1.11	common.logger.write(cmd)
258	slacapra	1.29	pass
259	slacapra	1.11
260	slacapra	1.16	child = popen2.Popen3(cmd, 1) # capture stdout and stderr from command
261			child.tochild.close() # don't need to talk to child
262			outfile = child.fromchild
263			outfd = outfile.fileno()
264			errfile = child.childerr
265			errfd = errfile.fileno()
266			makeNonBlocking(outfd) # don't deadlock!
267			makeNonBlocking(errfd)
268			outdata = []
269			errdata = []
270			outeof = erreof = 0
271
272			if timeout > 0 :
273			maxwaittime = time.time() + timeout
274
275			err = -1
276			while (timeout == -1 or time.time() < maxwaittime):
277			ready = select.select([outfd,errfd],[],[]) # wait for input
278			if outfd in ready[0]:
279			outchunk = outfile.read()
280			if outchunk == '': outeof = 1
281			outdata.append(outchunk)
282			if errfd in ready[0]:
283			errchunk = errfile.read()
284			if errchunk == '': erreof = 1
285			errdata.append(errchunk)
286			if outeof and erreof:
287			err = child.wait()
288			break
289	slacapra	1.21	select.select([],[],[],.01) # give a little time for buffers to fill
290	slacapra	1.16	if err == -1:
291			# kill the pid
292	slacapra	1.18	common.logger.message('killing process '+(cmd)+' with timeout '+str(timeout))
293	slacapra	1.16	os.kill (child.pid, 9)
294	slacapra	1.11	err = child.wait()
295	slacapra	1.16
296	fanzago	1.17	cmd_out = string.join(outdata,"")
297			cmd_err = string.join(errdata,"")
298
299	slacapra	1.11	if err:
300			common.logger.message('`'+cmd+'`\n failed with exit code '
301			+`err`+'='+`(err&0xff)`+'(signal)+'
302			+`(err>>8)`+'(status)')
303			common.logger.message(cmd_out)
304			common.logger.message(cmd_err)
305	nsmirnov	1.1	return None
306
307	fanzago	1.17	# cmd_out = string.join(outdata,"")
308			# cmd_err = string.join(errdata,"")
309	nsmirnov	1.1	cmd_out = cmd_out + cmd_err
310	slacapra	1.11	if printout:
311			common.logger.message(cmd_out)
312			else:
313	slacapra	1.13	common.logger.debug(10,cmd_out)
314	slacapra	1.11	common.logger.write(cmd_out)
315	slacapra	1.29	pass
316	slacapra	1.16	#print "<"+cmd_out+">"
317	nsmirnov	1.1	return cmd_out
318	nsmirnov	1.4
319	slacapra	1.11	####################################
320	gutsche	1.20	def makeCksum(filename) :
321			"""
322			make chksum using filename and content of file
323			"""
324
325	slacapra	1.31	tobedeleted=0
326	slacapra	1.29	try:
327			import tempfile
328			tmpfile= tempfile.NamedTemporaryFile(mode='w')
329			tmp_filename = tmpfile.name
330			except:
331			## SL in case py2.2 is used, fall back to old solution
332			tmp_filename = tempfile.mktemp()
333			tmpfile= open(tmp_filename,'w')
334			tobedeleted=1
335	gutsche	1.20
336			# add filename as first line
337	slacapra	1.29	tmpfile.write(filename+'\n')
338	gutsche	1.20
339			# fill input file in tmp file
340	slacapra	1.30	infile = open(filename)
341			tmpfile.writelines(infile.readlines())
342	slacapra	1.29	tmpfile.flush()
343	slacapra	1.30	infile.close()
344	gutsche	1.20
345			cmd = 'cksum '+tmp_filename
346			cmd_out = runCommand(cmd)
347
348			cksum = cmd_out.split()[0]
349
350	slacapra	1.29	## SL this is not needed if we use NamedTemporaryFile, which is not available in py2.2
351			if (tobedeleted): os.remove(tmp_filename)
352	gutsche	1.20
353			return cksum
354
355	gutsche	1.32	def spanRanges(jobArray):
356			"""
357			take array of job numbers and concatenate 1,2,3 to 1-3
358			return string
359			"""
360
361			output = ""
362	mcinquil	1.35	jobArray.sort()
363
364	gutsche	1.32	previous = jobArray[0]-1
365			for job in jobArray:
366			if previous+1 == job:
367			previous = job
368			if len(output) > 0 :
369			if output[-1] != "-":
370			output += "-"
371			else :
372			output += str(previous)
373			else:
374			output += str(previous) + "," + str(job)
375	mcinquil	1.35	#output += "," + str(job)
376	gutsche	1.32	previous = job
377			if len(jobArray) > 1 :
378			output += str(previous)
379
380			return output
381
382	gutsche	1.20	####################################
383	nsmirnov	1.4	if __name__ == '__main__':
384			print 'sys.argv[1] =',sys.argv[1]
385			list = parseRange2(sys.argv[1])
386			print list
387	slacapra	1.29	cksum = makeCksum("crab_util.py")
388			print cksum
389	nsmirnov	1.4