File: parsing.py

package info (click to toggle)
bootchart2 0.14.4-3
  • links: PTS, VCS
  • area: main
  • in suites: jessie, jessie-kfreebsd, stretch, wheezy
  • size: 5,036 kB
  • sloc: ansic: 1,805; python: 1,764; sh: 155; makefile: 88
file content (686 lines) | stat: -rw-r--r-- 26,086 bytes parent folder | download | duplicates (2)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
#  This file is part of pybootchartgui.

#  pybootchartgui is free software: you can redistribute it and/or modify
#  it under the terms of the GNU General Public License as published by
#  the Free Software Foundation, either version 3 of the License, or
#  (at your option) any later version.

#  pybootchartgui is distributed in the hope that it will be useful,
#  but WITHOUT ANY WARRANTY; without even the implied warranty of
#  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
#  GNU General Public License for more details.

#  You should have received a copy of the GNU General Public License
#  along with pybootchartgui. If not, see <http://www.gnu.org/licenses/>.


from __future__ import with_statement

import os
import string
import re
import tarfile
from time import clock
from collections import defaultdict
from functools import reduce

from .samples import *
from .process_tree import ProcessTree

# Parsing produces as its end result a 'Trace'

class Trace:
    def __init__(self, writer, paths, options):
        self.headers = None
        self.disk_stats = None
        self.ps_stats = None
        self.taskstats = None
        self.cpu_stats = None
        self.cmdline = None
        self.kernel = None
        self.kernel_tree = None
        self.filename = None
        self.parent_map = None
        self.mem_stats = None

        parse_paths (writer, self, paths)
        if not self.valid():
            raise ParseError("empty state: '%s' does not contain a valid bootchart" % ", ".join(paths))

        # Turn that parsed information into something more useful
        # link processes into a tree of pointers, calculate statistics
        self.compile(writer)

        # Crop the chart to the end of the first idle period after the given
        # process
        if options.crop_after:
            idle = self.crop (writer, options.crop_after)
        else:
            idle = None

        # Annotate other times as the first start point of given process lists
        self.times = [ idle ]
        if options.annotate:
            for procnames in options.annotate:
                names = [x[:15] for x in procnames.split(",")]
                for proc in self.ps_stats.process_map.values():
                    if proc.cmd in names:
                        self.times.append(proc.start_time)
                        break
                    else:
                        self.times.append(None)

        self.proc_tree = ProcessTree(writer, self.kernel, self.ps_stats,
                                     self.ps_stats.sample_period,
                                     self.headers.get("profile.process"),
                                     options.prune, idle, self.taskstats,
                                     self.parent_map is not None)

        if self.kernel is not None:
            self.kernel_tree = ProcessTree(writer, self.kernel, None, 0,
                                           self.headers.get("profile.process"),
                                           False, None, None, True)

    def valid(self):
        return self.headers != None and self.disk_stats != None and \
               self.ps_stats != None and self.cpu_stats != None


    def compile(self, writer):

        def find_parent_id_for(pid):
            if pid is 0:
                return 0
            ppid = self.parent_map.get(pid)
            if ppid:
                # many of these double forks are so short lived
                # that we have no samples, or process info for them
                # so climb the parent hierarcy to find one
                if int (ppid * 1000) not in self.ps_stats.process_map:
#                    print "Pid '%d' short lived with no process" % ppid
                    ppid = find_parent_id_for (ppid)
#                else:
#                    print "Pid '%d' has an entry" % ppid
            else:
#                print "Pid '%d' missing from pid map" % pid
                return 0
            return ppid

        # merge in the cmdline data
        if self.cmdline is not None:
            for proc in self.ps_stats.process_map.values():
                rpid = int (proc.pid / 1000)
                if rpid in self.cmdline:
                    cmd = self.cmdline[rpid]
                    proc.exe = cmd['exe']
                    proc.args = cmd['args']
#                else:
#                    print "proc %d '%s' not in cmdline" % (rpid, proc.exe)

        # re-parent any stray orphans if we can
        if self.parent_map is not None:
            for process in self.ps_stats.process_map.values():
                ppid = find_parent_id_for (int(process.pid / 1000))
                if ppid:
                    process.ppid = ppid * 1000

        # stitch the tree together with pointers
        for process in self.ps_stats.process_map.values():
            process.set_parent (self.ps_stats.process_map)

        # count on fingers variously
        for process in self.ps_stats.process_map.values():
            process.calc_stats (self.ps_stats.sample_period)

    def crop(self, writer, crop_after):

        def is_idle_at(util, start, j):
            k = j + 1
            while k < len(util) and util[k][0] < start + 300:
                k += 1
            k = min(k, len(util)-1)

            if util[j][1] >= 0.25:
                return False

            avgload = sum(u[1] for u in util[j:k+1]) / (k-j+1)
            if avgload < 0.25:
                return True
            else:
                return False
        def is_idle(util, start):
            for j in range(0, len(util)):
                if util[j][0] < start:
                    continue
                return is_idle_at(util, start, j)
            else:
                return False

        names = [x[:15] for x in crop_after.split(",")]
        for proc in self.ps_stats.process_map.values():
            if proc.cmd in names or proc.exe in names:
                writer.info("selected proc '%s' from list (start %d)"
                            % (proc.cmd, proc.start_time))
                break
        if proc is None:
            writer.warn("no selected crop proc '%s' in list" % crop_after)


        cpu_util = [(sample.time, sample.user + sample.sys + sample.io) for sample in self.cpu_stats]
        disk_util = [(sample.time, sample.util) for sample in self.disk_stats]

        idle = None
        for i in range(0, len(cpu_util)):
            if cpu_util[i][0] < proc.start_time:
                continue
            if is_idle_at(cpu_util, cpu_util[i][0], i) \
               and is_idle(disk_util, cpu_util[i][0]):
                idle = cpu_util[i][0]
                break

        if idle is None:
            writer.warn ("not idle after proc '%s'" % crop_after)
            return None

        crop_at = idle + 300
        writer.info ("cropping at time %d" % crop_at)
        while len (self.cpu_stats) \
                    and self.cpu_stats[-1].time > crop_at:
            self.cpu_stats.pop()
        while len (self.disk_stats) \
                    and self.disk_stats[-1].time > crop_at:
            self.disk_stats.pop()

        self.ps_stats.end_time = crop_at

        cropped_map = {}
        for key, value in self.ps_stats.process_map.items():
            if (value.start_time <= crop_at):
                cropped_map[key] = value

        for proc in cropped_map.values():
            proc.duration = min (proc.duration, crop_at - proc.start_time)
            while len (proc.samples) \
                        and proc.samples[-1].time > crop_at:
                proc.samples.pop()

        self.ps_stats.process_map = cropped_map

        return idle



class ParseError(Exception):
    """Represents errors during parse of the bootchart."""
    def __init__(self, value):
        self.value = value

    def __str__(self):
        return self.value

def _parse_headers(file):
    """Parses the headers of the bootchart."""
    def parse(acc, line):
        (headers, last) = acc
        if '=' in line:
            last, value = map (lambda x: x.strip(), line.split('=', 1))
        else:
            value = line.strip()
        headers[last] += value
        return headers, last
    return reduce(parse, file.read().decode('utf-8').split('\n'), (defaultdict(str),''))[0]

def _parse_timed_blocks(file):
    """Parses (ie., splits) a file into so-called timed-blocks. A
    timed-block consists of a timestamp on a line by itself followed
    by zero or more lines of data for that point in time."""
    def parse(block):
        lines = block.split('\n')
        if not lines:
            raise ParseError('expected a timed-block consisting a timestamp followed by data lines')
        try:
            return (int(lines[0]), lines[1:])
        except ValueError:
            raise ParseError("expected a timed-block, but timestamp '%s' is not an integer" % lines[0])
    blocks = file.read().decode('utf-8').split('\n\n')
    return [parse(block) for block in blocks if block.strip() and not block.endswith(' not running\n')]

def _parse_proc_ps_log(writer, file):
    """
     * See proc(5) for details.
     *
     * {pid, comm, state, ppid, pgrp, session, tty_nr, tpgid, flags, minflt, cminflt, majflt, cmajflt, utime, stime,
     *  cutime, cstime, priority, nice, 0, itrealvalue, starttime, vsize, rss, rlim, startcode, endcode, startstack,
     *  kstkesp, kstkeip}
    """
    processMap = {}
    ltime = 0
    timed_blocks = _parse_timed_blocks(file)
    for time, lines in timed_blocks:
        for line in lines:
            if line is '': continue
            tokens = line.split(' ')
            if len(tokens) < 21:
                continue

            offset = [index for index, token in enumerate(tokens[1:]) if token[-1] == ')'][0]
            pid, cmd, state, ppid = int(tokens[0]), ' '.join(tokens[1:2+offset]), tokens[2+offset], int(tokens[3+offset])
            userCpu, sysCpu, stime = int(tokens[13+offset]), int(tokens[14+offset]), int(tokens[21+offset])

            # magic fixed point-ness ...
            pid *= 1000
            ppid *= 1000
            if pid in processMap:
                process = processMap[pid]
                process.cmd = cmd.strip('()') # why rename after latest name??
            else:
                process = Process(writer, pid, cmd.strip('()'), ppid, min(time, stime))
                processMap[pid] = process

            if process.last_user_cpu_time is not None and process.last_sys_cpu_time is not None and ltime is not None:
                userCpuLoad, sysCpuLoad = process.calc_load(userCpu, sysCpu, max(1, time - ltime))
                cpuSample = CPUSample('null', userCpuLoad, sysCpuLoad, 0.0)
                process.samples.append(ProcessSample(time, state, cpuSample))

            process.last_user_cpu_time = userCpu
            process.last_sys_cpu_time = sysCpu
        ltime = time

    if len (timed_blocks) < 2:
        return None

    startTime = timed_blocks[0][0]
    avgSampleLength = (ltime - startTime)/(len (timed_blocks) - 1)

    return ProcessStats (writer, processMap, len (timed_blocks), avgSampleLength, startTime, ltime)

def _parse_taskstats_log(writer, file):
    """
     * See bootchart-collector.c for details.
     *
     * { pid, ppid, comm, cpu_run_real_total, blkio_delay_total, swapin_delay_total }
     *
    """
    processMap = {}
    pidRewrites = {}
    ltime = None
    timed_blocks = _parse_timed_blocks(file)
    for time, lines in timed_blocks:
        # we have no 'stime' from taskstats, so prep 'init'
        if ltime is None:
            process = Process(writer, 1, '[init]', 0, 0)
            processMap[1000] = process
            ltime = time
#                       continue
        for line in lines:
            if line is '': continue
            tokens = line.split(' ')
            if len(tokens) != 6:
                continue

            opid, ppid, cmd = int(tokens[0]), int(tokens[1]), tokens[2]
            try:
                cpu_ns, blkio_delay_ns, swapin_delay_ns = long(tokens[-3]), long(tokens[-2]), long(tokens[-1]),
            except NameError:
                cpu_ns, blkio_delay_ns, swapin_delay_ns = int(tokens[-3]), int(tokens[-2]), int(tokens[-1]),

            # make space for trees of pids
            opid *= 1000
            ppid *= 1000

            # when the process name changes, we re-write the pid.
            if opid in pidRewrites:
                pid = pidRewrites[opid]
            else:
                pid = opid

            cmd = cmd.strip('(').strip(')')
            if pid in processMap:
                process = processMap[pid]
                if process.cmd != cmd:
                    pid += 1
                    pidRewrites[opid] = pid
#                                       print "process mutation ! '%s' vs '%s' pid %s -> pid %s\n" % (process.cmd, cmd, opid, pid)
                    process = process.split (writer, pid, cmd, ppid, time)
                    processMap[pid] = process
                else:
                    process.cmd = cmd;
            else:
                process = Process(writer, pid, cmd, ppid, time)
                processMap[pid] = process

            delta_cpu_ns = (float) (cpu_ns - process.last_cpu_ns)
            delta_blkio_delay_ns = (float) (blkio_delay_ns - process.last_blkio_delay_ns)
            delta_swapin_delay_ns = (float) (swapin_delay_ns - process.last_swapin_delay_ns)

            # make up some state data ...
            if delta_cpu_ns > 0:
                state = "R"
            elif delta_blkio_delay_ns + delta_swapin_delay_ns > 0:
                state = "D"
            else:
                state = "S"

            # retain the ns timing information into a CPUSample - that tries
            # with the old-style to be a %age of CPU used in this time-slice.
            if delta_cpu_ns + delta_blkio_delay_ns + delta_swapin_delay_ns > 0:
#                               print "proc %s cpu_ns %g delta_cpu %g" % (cmd, cpu_ns, delta_cpu_ns)
                cpuSample = CPUSample('null', delta_cpu_ns, 0.0,
                                      delta_blkio_delay_ns,
                                      delta_swapin_delay_ns)
                process.samples.append(ProcessSample(time, state, cpuSample))

            process.last_cpu_ns = cpu_ns
            process.last_blkio_delay_ns = blkio_delay_ns
            process.last_swapin_delay_ns = swapin_delay_ns
        ltime = time

    if len (timed_blocks) < 2:
        return None

    startTime = timed_blocks[0][0]
    avgSampleLength = (ltime - startTime)/(len(timed_blocks)-1)

    return ProcessStats (writer, processMap, len (timed_blocks), avgSampleLength, startTime, ltime)

def _parse_proc_stat_log(file):
    samples = []
    ltimes = None
    for time, lines in _parse_timed_blocks(file):
        # skip emtpy lines
        if not lines:
            continue
        # CPU times {user, nice, system, idle, io_wait, irq, softirq}
        tokens = lines[0].split()
        times = [ int(token) for token in tokens[1:] ]
        if ltimes:
            user = float((times[0] + times[1]) - (ltimes[0] + ltimes[1]))
            system = float((times[2] + times[5] + times[6]) - (ltimes[2] + ltimes[5] + ltimes[6]))
            idle = float(times[3] - ltimes[3])
            iowait = float(times[4] - ltimes[4])

            aSum = max(user + system + idle + iowait, 1)
            samples.append( CPUSample(time, user/aSum, system/aSum, iowait/aSum) )

        ltimes = times
        # skip the rest of statistics lines
    return samples

def _parse_proc_disk_stat_log(file, numCpu):
    """
    Parse file for disk stats, but only look at the whole device, eg. sda,
    not sda1, sda2 etc. The format of relevant lines should be:
    {major minor name rio rmerge rsect ruse wio wmerge wsect wuse running use aveq}
    """
    disk_regex_re = re.compile ('^([hsv]d.|mtdblock\d|mmcblk\d|cciss/c\d+d\d+.*)$')

    # this gets called an awful lot.
    def is_relevant_line(linetokens):
        if len(linetokens) != 14:
            return False
        disk = linetokens[2]
        return disk_regex_re.match(disk)

    disk_stat_samples = []

    for time, lines in _parse_timed_blocks(file):
        sample = DiskStatSample(time)
        relevant_tokens = [linetokens for linetokens in map (lambda x: x.split(),lines) if is_relevant_line(linetokens)]

        for tokens in relevant_tokens:
            disk, rsect, wsect, use = tokens[2], int(tokens[5]), int(tokens[9]), int(tokens[12])
            sample.add_diskdata([rsect, wsect, use])

        disk_stat_samples.append(sample)

    disk_stats = []
    for sample1, sample2 in zip(disk_stat_samples[:-1], disk_stat_samples[1:]):
        interval = sample1.time - sample2.time
        if interval == 0:
            interval = 1
        sums = [ a - b for a, b in zip(sample1.diskdata, sample2.diskdata) ]
        readTput = sums[0] / 2.0 * 100.0 / interval
        writeTput = sums[1] / 2.0 * 100.0 / interval
        util = float( sums[2] ) / 10 / interval / numCpu
        util = max(0.0, min(1.0, util))
        disk_stats.append(DiskSample(sample2.time, readTput, writeTput, util))

    return disk_stats

def _parse_proc_meminfo_log(file):
    """
    Parse file for global memory statistics.
    The format of relevant lines should be: ^key: value( unit)?
    """
    mem_stats = []
    meminfo_re = re.compile(r'([^ \t:]+):\s*(\d+).*')

    for time, lines in _parse_timed_blocks(file):
        sample = MemSample(time)

        for line in lines:
            match = meminfo_re.match(line)
            if match:
                sample.add_value(match.group(1), int(match.group(2)))

        if sample.valid():
            mem_stats.append(sample)

    return mem_stats

# if we boot the kernel with: initcall_debug printk.time=1 we can
# get all manner of interesting data from the dmesg output
# We turn this into a pseudo-process tree: each event is
# characterised by a
# we don't try to detect a "kernel finished" state - since the kernel
# continues to do interesting things after init is called.
#
# sample input:
# [    0.000000] ACPI: FACP 3f4fc000 000F4 (v04 INTEL  Napa     00000001 MSFT 01000013)
# ...
# [    0.039993] calling  migration_init+0x0/0x6b @ 1
# [    0.039993] initcall migration_init+0x0/0x6b returned 1 after 0 usecs
def _parse_dmesg(writer, file):
    timestamp_re = re.compile ("^\[\s*(\d+\.\d+)\s*]\s+(.*)$")
    split_re = re.compile ("^(\S+)\s+([\S\+_-]+) (.*)$")
    processMap = {}
    idx = 0
    inc = 1.0 / 1000000
    kernel = Process(writer, idx, "k-boot", 0, 0.1)
    processMap['k-boot'] = kernel
    base_ts = False
    max_ts = 0
    for line in file.read().decode('utf-8').split('\n'):
        t = timestamp_re.match (line)
        if t is None:
#                       print "duff timestamp " + line
            continue

        time_ms = float (t.group(1)) * 1000
        # looks like we may have a huge diff after the clock
        # has been set up. This could lead to huge graph:
        # so huge we will be killed by the OOM.
        # So instead of using the plain timestamp we will
        # use a delta to first one and skip the first one
        # for convenience
        if max_ts == 0 and not base_ts and time_ms > 1000:
            base_ts = time_ms
            continue
        max_ts = max(time_ms, max_ts)
        if base_ts:
#                       print "fscked clock: used %f instead of %f" % (time_ms - base_ts, time_ms)
            time_ms -= base_ts
        m = split_re.match (t.group(2))

        if m is None:
            continue
#               print "match: '%s'" % (m.group(1))
        type = m.group(1)
        func = m.group(2)
        rest = m.group(3)

        if t.group(2).startswith ('Write protecting the') or \
           t.group(2).startswith ('Freeing unused kernel memory'):
            kernel.duration = time_ms / 10
            continue

#               print "foo: '%s' '%s' '%s'" % (type, func, rest)
        if type == "calling":
            ppid = kernel.pid
            p = re.match ("\@ (\d+)", rest)
            if p is not None:
                ppid = float (p.group(1)) / 1000
#                               print "match: '%s' ('%g') at '%s'" % (func, ppid, time_ms)
            name = func.split ('+', 1) [0]
            idx += inc
            processMap[func] = Process(writer, ppid + idx, name, ppid, time_ms / 10)
        elif type == "initcall":
#                       print "finished: '%s' at '%s'" % (func, time_ms)
            if func in processMap:
                process = processMap[func]
                process.duration = (time_ms / 10) - process.start_time
            else:
                print("corrupted init call for %s" % (func))

        elif type == "async_waiting" or type == "async_continuing":
            continue # ignore

    return processMap.values()

#
# Parse binary pacct accounting file output if we have one
# cf. /usr/include/linux/acct.h
#
def _parse_pacct(writer, file):
    # read LE int32
    def _read_le_int32(file):
        byts = file.read(4)
        return (ord(byts[0]))       | (ord(byts[1]) << 8) | \
               (ord(byts[2]) << 16) | (ord(byts[3]) << 24)

    parent_map = {}
    parent_map[0] = 0
    while file.read(1) != "": # ignore flags
        ver = file.read(1)
        if ord(ver) < 3:
            print("Invalid version 0x%x" % (ord(ver)))
            return None

        file.seek (14, 1)     # user, group etc.
        pid = _read_le_int32 (file)
        ppid = _read_le_int32 (file)
#               print "Parent of %d is %d" % (pid, ppid)
        parent_map[pid] = ppid
        file.seek (4 + 4 + 16, 1) # timings
        file.seek (16, 1)         # acct_comm
    return parent_map

def _parse_paternity_log(writer, file):
    parent_map = {}
    parent_map[0] = 0
    for line in file.read().decode('utf-8').split('\n'):
        elems = line.split(' ') # <Child> <Parent>
        if len (elems) >= 2:
#                       print "paternity of %d is %d" % (int(elems[0]), int(elems[1]))
            parent_map[int(elems[0])] = int(elems[1])
        elif line is not '':
            print("Odd paternity line '%s'" % (line))
    return parent_map

def _parse_cmdline_log(writer, file):
    cmdLines = {}
    for block in file.read().decode('utf-8').split('\n\n'):
        lines = block.split('\n')
        if len (lines) >= 3:
#                       print "Lines '%s'" % (lines[0])
            pid = int (lines[0])
            values = {}
            values['exe'] = lines[1].lstrip(':')
            args = lines[2].lstrip(':').split('\0')
            args.pop()
            values['args'] = args
            cmdLines[pid] = values
    return cmdLines

def get_num_cpus(headers):
    """Get the number of CPUs from the system.cpu header property. As the
    CPU utilization graphs are relative, the number of CPUs currently makes
    no difference."""
    if headers is None:
        return 1
    if headers.get("system.cpu.num"):
        return max (int (headers.get("system.cpu.num")), 1)
    cpu_model = headers.get("system.cpu")
    if cpu_model is None:
        return 1
    mat = re.match(".*\\((\\d+)\\)", cpu_model)
    if mat is None:
        return 1
    return max (int(mat.group(1)), 1)

def _do_parse(writer, state, name, file):
    writer.status("parsing '%s'" % name)
    t1 = clock()
    if name == "header":
        state.headers = _parse_headers(file)
    elif name == "proc_diskstats.log":
        state.disk_stats = _parse_proc_disk_stat_log(file, get_num_cpus(state.headers))
    elif name == "taskstats.log":
        state.ps_stats = _parse_taskstats_log(writer, file)
        state.taskstats = True
    elif name == "proc_stat.log":
        state.cpu_stats = _parse_proc_stat_log(file)
    elif name == "proc_meminfo.log":
        state.mem_stats = _parse_proc_meminfo_log(file)
    elif name == "dmesg":
        state.kernel = _parse_dmesg(writer, file)
    elif name == "cmdline2.log":
        state.cmdline = _parse_cmdline_log(writer, file)
    elif name == "paternity.log":
        state.parent_map = _parse_paternity_log(writer, file)
    elif name == "proc_ps.log":  # obsoleted by TASKSTATS
        state.ps_stats = _parse_proc_ps_log(writer, file)
    elif name == "kernel_pacct": # obsoleted by PROC_EVENTS
        state.parent_map = _parse_pacct(writer, file)
    t2 = clock()
    writer.info("  %s seconds" % str(t2-t1))
    return state

def parse_file(writer, state, filename):
    if state.filename is None:
        state.filename = filename
    basename = os.path.basename(filename)
    with open(filename, "rb") as file:
        return _do_parse(writer, state, basename, file)

def parse_paths(writer, state, paths):
    for path in paths:
        root, extension = os.path.splitext(path)
        if not(os.path.exists(path)):
            writer.warn("warning: path '%s' does not exist, ignoring." % path)
            continue
        state.filename = path
        if os.path.isdir(path):
            files = [ f for f in [os.path.join(path, f) for f in os.listdir(path)] if os.path.isfile(f) ]
            files.sort()
            state = parse_paths(writer, state, files)
        elif extension in [".tar", ".tgz", ".gz"]:
            if extension == ".gz":
                root, extension = os.path.splitext(root)
                if extension != ".tar":
                    writer.warn("warning: can only handle zipped tar files, not zipped '%s'-files; ignoring" % extension)
                    continue
            tf = None
            try:
                writer.status("parsing '%s'" % path)
                tf = tarfile.open(path, 'r:*')
                for name in tf.getnames():
                    state = _do_parse(writer, state, name, tf.extractfile(name))
            except tarfile.ReadError as error:
                raise ParseError("error: could not read tarfile '%s': %s." % (path, error))
            finally:
                if tf != None:
                    tf.close()
        else:
            state = parse_file(writer, state, path)
    return state