1# http://pyrocko.org - GPLv3
2#
3# The Pyrocko Developers, 21st Century
4# ---|P------/S----------~Lg----------
6from __future__ import absolute_import, print_function
8import re
9import logging
10import os.path as op
12import numpy as num
14from pyrocko import io, trace, util
15from pyrocko.progress import progress
16from pyrocko.has_paths import Path, HasPaths
17from pyrocko.guts import Dict, String, Choice, Float, List, Timestamp, \
18 StringChoice, IntChoice, Defer, load_all, clone
19from pyrocko.squirrel.dataset import Dataset
20from pyrocko.squirrel.client.local import LocalData
21from pyrocko.squirrel.error import ToolError
22from pyrocko.squirrel.model import CodesNSLCE
23from pyrocko.squirrel.operators.base import NetworkGrouping, StationGrouping, \
24 ChannelGrouping, SensorGrouping
26tts = util.time_to_str
28guts_prefix = 'jackseis'
29logger = logging.getLogger('psq.cli.jackseis')
32def make_task(*args):
33 return progress.task(*args, logger=logger)
36class JackseisError(ToolError):
37 pass
40class Chain(object):
41 def __init__(self, node, parent=None):
42 self.node = node
43 self.parent = parent
45 def mcall(self, name, *args, **kwargs):
46 ret = []
47 if self.parent is not None:
48 ret.append(self.parent.mcall(name, *args, **kwargs))
50 ret.append(getattr(self.node, name)(*args, **kwargs))
51 return ret
53 def fcall(self, name, *args, **kwargs):
54 v = getattr(self.node, name)(*args, **kwargs)
55 if v is None and self.parent is not None:
56 return self.parent.fcall(name, *args, **kwargs)
57 else:
58 return v
60 def get(self, name):
61 v = getattr(self.node, name)
62 if v is None and self.parent is not None:
63 return self.parent.get(name)
64 else:
65 return v
67 def dget(self, name, k):
68 v = getattr(self.node, name).get(k, None)
69 if v is None and self.parent is not None:
70 return self.parent.dget(name, k)
71 else:
72 return v
75class OutputFormatChoice(StringChoice):
76 choices = io.allowed_formats('save')
79class OutputDataTypeChoice(StringChoice):
80 choices = ['int32', 'int64', 'float32', 'float64']
81 name_to_dtype = {
82 'int32': num.int32,
83 'int64': num.int64,
84 'float32': num.float32,
85 'float64': num.float64}
88class TraversalChoice(StringChoice):
89 choices = ['network', 'station', 'channel', 'sensor']
90 name_to_grouping = {
91 'network': NetworkGrouping(),
92 'station': StationGrouping(),
93 'sensor': SensorGrouping(),
94 'channel': ChannelGrouping()}
97class Converter(HasPaths):
99 in_dataset = Dataset.T(optional=True)
100 in_path = String.T(optional=True)
101 in_paths = List.T(String.T(optional=True))
103 codes = List.T(CodesNSLCE.T(), optional=True)
105 rename = Dict.T(
106 String.T(),
107 Choice.T([
108 String.T(),
109 Dict.T(String.T(), String.T())]))
110 tmin = Timestamp.T(optional=True)
111 tmax = Timestamp.T(optional=True)
112 tinc = Float.T(optional=True)
114 downsample = Float.T(optional=True)
116 out_path = Path.T(optional=True)
117 out_sds_path = Path.T(optional=True)
118 out_format = OutputFormatChoice.T(optional=True)
119 out_data_type = OutputDataTypeChoice.T(optional=True)
120 out_mseed_record_length = IntChoice.T(
121 optional=True,
122 choices=list(io.mseed.VALID_RECORD_LENGTHS))
123 out_mseed_steim = IntChoice.T(
124 optional=True,
125 choices=[1, 2])
126 out_meta_path = Path.T(optional=True)
128 traversal = TraversalChoice.T(optional=True)
130 parts = List.T(Defer('Converter.T'))
132 @classmethod
133 def add_arguments(cls, p):
134 p.add_squirrel_query_arguments(without=['time'])
136 p.add_argument(
137 '--downsample',
138 dest='downsample',
139 type=float,
140 metavar='RATE',
141 help='Downsample to RATE [Hz].')
143 p.add_argument(
144 '--out-path',
145 dest='out_path',
146 metavar='TEMPLATE',
147 help='Set output path to TEMPLATE. Available placeholders '
148 'are %%n: network, %%s: station, %%l: location, %%c: '
149 'channel, %%b: begin time, %%e: end time, %%j: julian day of '
150 'year. The following additional placeholders use the window '
151 'begin and end times rather than trace begin and end times '
152 '(to suppress producing many small files for gappy traces), '
153 '%%(wmin_year)s, %%(wmin_month)s, %%(wmin_day)s, %%(wmin)s, '
154 '%%(wmin_jday)s, %%(wmax_year)s, %%(wmax_month)s, '
155 '%%(wmax_day)s, %%(wmax)s, %%(wmax_jday)s. '
156 'Example: --output=\'data/%%s/trace-%%s-%%c.mseed\'')
158 p.add_argument(
159 '--out-sds-path',
160 dest='out_sds_path',
161 metavar='PATH',
162 help='Set output path to create SDS (https://www.seiscomp.de'
163 '/seiscomp3/doc/applications/slarchive/SDS.html), rooted at '
164 'the given path. Implies --tinc=86400. '
165 'Example: --output-sds-path=data/sds')
167 p.add_argument(
168 '--out-format',
169 dest='out_format',
170 choices=io.allowed_formats('save'),
171 metavar='FORMAT',
172 help='Set output file format. Choices: %s' % io.allowed_formats(
173 'save', 'cli_help', 'mseed'))
175 p.add_argument(
176 '--out-data-type',
177 dest='out_data_type',
178 choices=OutputDataTypeChoice.choices,
179 metavar='DTYPE',
180 help='Set numerical data type. Choices: %s. The output file '
181 'format must support the given type. By default, the data '
182 'type is kept unchanged.' % ', '.join(
183 OutputDataTypeChoice.choices))
185 p.add_argument(
186 '--out-mseed-record-length',
187 dest='out_mseed_record_length',
188 type=int,
189 choices=io.mseed.VALID_RECORD_LENGTHS,
190 metavar='INT',
191 help='Set the mseed record length in bytes. Choices: %s. '
192 'Default is 4096 bytes, which is commonly used for archiving.'
193 % ', '.join(str(b) for b in io.mseed.VALID_RECORD_LENGTHS))
195 p.add_argument(
196 '--out-mseed-steim',
197 dest='out_mseed_steim',
198 type=int,
199 choices=(1, 2),
200 metavar='INT',
201 help='Set the mseed STEIM compression. Choices: 1 or 2. '
202 'Default is STEIM-2, which can compress full range int32. '
203 'Note: STEIM-2 is limited to 30 bit dynamic range.')
205 p.add_argument(
206 '--out-meta-path',
207 dest='out_meta_path',
208 metavar='PATH',
209 help='Set output path for station metadata (StationXML) export.')
211 p.add_argument(
212 '--traversal',
213 dest='traversal',
214 metavar='GROUPING',
215 choices=TraversalChoice.choices,
216 help='By default the outermost processing loop is over time. '
217 'Add outer loop with given GROUPING. Choices: %s'
218 % ', '.join(TraversalChoice.choices))
220 @classmethod
221 def from_arguments(cls, args):
222 kwargs = args.squirrel_query
224 obj = cls(
225 downsample=args.downsample,
226 out_format=args.out_format,
227 out_path=args.out_path,
228 out_sds_path=args.out_sds_path,
229 out_data_type=args.out_data_type,
230 out_mseed_record_length=args.out_mseed_record_length,
231 out_mseed_steim=args.out_mseed_steim,
232 out_meta_path=args.out_meta_path,
233 traversal=args.traversal,
234 **kwargs)
236 obj.validate()
237 return obj
239 def add_dataset(self, sq):
240 if self.in_dataset is not None:
241 sq.add_dataset(self.in_dataset)
243 if self.in_path is not None:
244 ds = Dataset(sources=[LocalData(paths=[self.in_path])])
245 ds.set_basepath_from(self)
246 sq.add_dataset(ds)
248 if self.in_paths:
249 ds = Dataset(sources=[LocalData(paths=self.in_paths)])
250 ds.set_basepath_from(self)
251 sq.add_dataset(ds)
253 def get_effective_rename_rules(self, chain):
254 d = {}
255 for k in ['network', 'station', 'location', 'channel']:
256 v = chain.dget('rename', k)
257 if isinstance(v, str):
258 m = re.match(r'/([^/]+)/([^/]*)/', v)
259 if m:
260 try:
261 v = (re.compile(m.group(1)), m.group(2))
262 except Exception:
263 raise JackseisError(
264 'Invalid replacement pattern: /%s/' % m.group(1))
266 d[k] = v
268 return d
270 def get_effective_out_path(self):
271 nset = sum(x is not None for x in (
272 self.out_path,
273 self.out_sds_path))
275 if nset > 1:
276 raise JackseisError(
277 'More than one out of [out_path, out_sds_path] set.')
279 is_sds = False
280 if self.out_path:
281 out_path = self.out_path
283 elif self.out_sds_path:
284 out_path = op.join(
285 self.out_sds_path,
286 '%(wmin_year)s/%(network)s/%(station)s/%(channel)s.D'
287 '/%(network)s.%(station)s.%(location)s.%(channel)s.D'
288 '.%(wmin_year)s.%(wmin_jday)s')
289 is_sds = True
290 else:
291 out_path = None
293 if out_path is not None:
294 return self.expand_path(out_path), is_sds
295 else:
296 return None
298 def get_effective_out_meta_path(self):
299 if self.out_meta_path is not None:
300 return self.expand_path(self.out_meta_path)
301 else:
302 return None
304 def do_rename(self, rules, tr):
305 rename = {}
306 for k in ['network', 'station', 'location', 'channel']:
307 v = rules.get(k, None)
308 if isinstance(v, str):
309 rename[k] = v
310 elif isinstance(v, dict):
311 try:
312 oldval = getattr(tr, k)
313 rename[k] = v[oldval]
314 except KeyError:
315 raise ToolError(
316 'No mapping defined for %s code "%s".' % (k, oldval))
318 elif isinstance(v, tuple):
319 pat, repl = v
320 oldval = getattr(tr, k)
321 newval, n = pat.subn(repl, oldval)
322 if n:
323 rename[k] = newval
325 tr.set_codes(**rename)
327 def convert(self, args, chain=None):
328 if chain is None:
329 defaults = clone(g_defaults)
330 defaults.set_basepath_from(self)
331 chain = Chain(defaults)
333 chain = Chain(self, chain)
335 if self.parts:
336 task = make_task('Jackseis parts')
337 for part in task(self.parts):
338 part.convert(args, chain)
340 del task
342 else:
343 sq = args.make_squirrel()
345 cli_overrides = Converter.from_arguments(args)
346 cli_overrides.set_basepath('.')
348 chain = Chain(cli_overrides, chain)
350 chain.mcall('add_dataset', sq)
352 tmin = chain.get('tmin')
353 tmax = chain.get('tmax')
354 tinc = chain.get('tinc')
355 codes = chain.get('codes')
356 downsample = chain.get('downsample')
357 out_path, is_sds = chain.fcall('get_effective_out_path') \
358 or (None, False)
360 if is_sds and tinc != 86400.:
361 logger.warning('Setting time window to 1 day to generate SDS.')
362 tinc = 86400.0
364 out_format = chain.get('out_format')
365 out_data_type = chain.get('out_data_type')
367 out_meta_path = chain.fcall('get_effective_out_meta_path')
369 if out_meta_path is not None:
370 sx = sq.get_stationxml(codes=codes, tmin=tmin, tmax=tmax)
371 util.ensuredirs(out_meta_path)
372 sx.dump_xml(filename=out_meta_path)
373 if out_path is None:
374 return
376 target_deltat = None
377 if downsample is not None:
378 target_deltat = 1.0 / float(downsample)
380 save_kwargs = {}
381 if out_format == 'mseed':
382 save_kwargs['record_length'] = chain.get(
383 'out_mseed_record_length')
384 save_kwargs['steim'] = chain.get(
385 'out_mseed_steim')
387 traversal = chain.get('traversal')
388 if traversal is not None:
389 grouping = TraversalChoice.name_to_grouping[traversal]
390 else:
391 grouping = None
393 tpad = 0.0
394 if target_deltat is not None:
395 tpad += target_deltat * 50.
397 task = None
398 rename_rules = self.get_effective_rename_rules(chain)
399 for batch in sq.chopper_waveforms(
400 tmin=tmin, tmax=tmax, tpad=tpad, tinc=tinc,
401 codes=codes,
402 snap_window=True,
403 grouping=grouping):
405 if task is None:
406 task = make_task(
407 'Jackseis blocks', batch.n * batch.ngroups)
409 tlabel = '%s%s - %s' % (
410 'groups %i / %i: ' % (batch.igroup, batch.ngroups)
411 if batch.ngroups > 1 else '',
412 util.time_to_str(batch.tmin),
413 util.time_to_str(batch.tmax))
415 task.update(batch.i + batch.igroup * batch.n, tlabel)
417 twmin = batch.tmin
418 twmax = batch.tmax
420 traces = batch.traces
422 if target_deltat is not None:
423 out_traces = []
424 for tr in traces:
425 try:
426 tr.downsample_to(
427 target_deltat, snap=True, demean=False)
429 out_traces.append(tr)
431 except (trace.TraceTooShort, trace.NoData):
432 pass
434 traces = out_traces
436 for tr in traces:
437 self.do_rename(rename_rules, tr)
439 if out_data_type:
440 for tr in traces:
441 tr.ydata = tr.ydata.astype(
442 OutputDataTypeChoice.name_to_dtype[out_data_type])
444 chopped_traces = []
445 for tr in traces:
446 try:
447 otr = tr.chop(twmin, twmax, inplace=False)
448 chopped_traces.append(otr)
449 except trace.NoData:
450 pass
452 traces = chopped_traces
454 if out_path is not None:
455 try:
456 io.save(
457 traces, out_path,
458 format=out_format,
459 overwrite=args.force,
460 additional=dict(
461 wmin_year=tts(twmin, format='%Y'),
462 wmin_month=tts(twmin, format='%m'),
463 wmin_day=tts(twmin, format='%d'),
464 wmin_jday=tts(twmin, format='%j'),
465 wmin=tts(twmin, format='%Y-%m-%d_%H-%M-%S'),
466 wmax_year=tts(twmax, format='%Y'),
467 wmax_month=tts(twmax, format='%m'),
468 wmax_day=tts(twmax, format='%d'),
469 wmax_jday=tts(twmax, format='%j'),
470 wmax=tts(twmax, format='%Y-%m-%d_%H-%M-%S')),
471 **save_kwargs)
473 except io.FileSaveError as e:
474 raise JackseisError(str(e))
476 else:
477 for tr in traces:
478 print(tr.summary)
480 if task:
481 task.done()
484g_defaults = Converter(
485 out_mseed_record_length=4096,
486 out_format='mseed',
487 out_mseed_steim=2)
490headline = 'Convert waveform archive data.'
493def make_subparser(subparsers):
494 return subparsers.add_parser(
495 'jackseis',
496 help=headline,
497 description=headline)
500def setup(parser):
501 parser.add_squirrel_selection_arguments()
503 parser.add_argument(
504 '--config',
505 dest='config_path',
506 metavar='NAME',
507 help='File containing `jackseis.Converter` settings.')
509 parser.add_argument(
510 '--force',
511 dest='force',
512 action='store_true',
513 default=False,
514 help='Force overwriting of existing files.')
516 Converter.add_arguments(parser)
519def run(parser, args):
520 if args.config_path:
521 try:
522 converters = load_all(filename=args.config_path)
523 except Exception as e:
524 raise ToolError(str(e))
526 for converter in converters:
527 if not isinstance(converter, Converter):
528 raise ToolError(
529 'Config file should only contain '
530 '`jackseis.Converter` objects.')
532 converter.set_basepath(op.dirname(args.config_path))
534 else:
535 converter = Converter()
536 converter.set_basepath('.')
537 converters = [converter]
539 with progress.view():
540 task = make_task('Jackseis jobs')
541 for converter in task(converters):
542 converter.convert(args)