1# http://pyrocko.org - GPLv3
2#
3# The Pyrocko Developers, 21st Century
4# ---|P------/S----------~Lg----------
6from __future__ import absolute_import, print_function
8import sys
9import os
11import math
12import logging
13import threading
14import queue
15from collections import defaultdict
17from pyrocko.guts import Object, Int, List, Tuple, String, Timestamp, Dict
18from pyrocko import util, trace
19from pyrocko.progress import progress
21from . import model, io, cache, dataset
23from .model import to_kind_id, WaveformOrder, to_kind, to_codes, \
24 STATION, CHANNEL, RESPONSE, EVENT, WAVEFORM, codes_patterns_list, \
25 codes_patterns_for_kind
26from .client import fdsn, catalog
27from .selection import Selection, filldocs
28from .database import abspath
29from .operators.base import Operator, CodesPatternFiltering
30from . import client, environment, error
32logger = logging.getLogger('psq.base')
34guts_prefix = 'squirrel'
37def make_task(*args):
38 return progress.task(*args, logger=logger)
41def lpick(condition, seq):
42 ft = [], []
43 for ele in seq:
44 ft[int(bool(condition(ele)))].append(ele)
46 return ft
49def len_plural(obj):
50 return len(obj), '' if len(obj) == 1 else 's'
53def blocks(tmin, tmax, deltat, nsamples_block=100000):
54 tblock = util.to_time_float(deltat * nsamples_block)
55 iblock_min = int(math.floor(tmin / tblock))
56 iblock_max = int(math.ceil(tmax / tblock))
57 for iblock in range(iblock_min, iblock_max):
58 yield iblock * tblock, (iblock+1) * tblock
61def gaps(avail, tmin, tmax):
62 assert tmin < tmax
64 data = [(tmax, 1), (tmin, -1)]
65 for (tmin_a, tmax_a) in avail:
66 assert tmin_a < tmax_a
67 data.append((tmin_a, 1))
68 data.append((tmax_a, -1))
70 data.sort()
71 s = 1
72 gaps = []
73 tmin_g = None
74 for t, x in data:
75 if s == 1 and x == -1:
76 tmin_g = t
77 elif s == 0 and x == 1 and tmin_g is not None:
78 tmax_g = t
79 if tmin_g != tmax_g:
80 gaps.append((tmin_g, tmax_g))
82 s += x
84 return gaps
87def order_key(order):
88 return (order.codes, order.tmin, order.tmax)
91def _is_exact(pat):
92 return not ('*' in pat or '?' in pat or ']' in pat or '[' in pat)
95def prefix_tree(tups):
96 if not tups:
97 return []
99 if len(tups[0]) == 1:
100 return sorted((tup[0], []) for tup in tups)
102 d = defaultdict(list)
103 for tup in tups:
104 d[tup[0]].append(tup[1:])
106 sub = []
107 for k in sorted(d.keys()):
108 sub.append((k, prefix_tree(d[k])))
110 return sub
113def match_time_span(tmin, tmax, obj):
114 return (obj.tmin is None or tmax is None or obj.tmin <= tmax) \
115 and (tmin is None or obj.tmax is None or tmin < obj.tmax)
118class Batch(object):
119 '''
120 Batch of waveforms from window-wise data extraction.
122 Encapsulates state and results yielded for each window in window-wise
123 waveform extraction with the :py:meth:`Squirrel.chopper_waveforms` method.
125 *Attributes:*
127 .. py:attribute:: tmin
129 Start of this time window.
131 .. py:attribute:: tmax
133 End of this time window.
135 .. py:attribute:: i
137 Index of this time window in sequence.
139 .. py:attribute:: n
141 Total number of time windows in sequence.
143 .. py:attribute:: igroup
145 Index of this time window's sequence group.
147 .. py:attribute:: ngroups
149 Total number of sequence groups.
151 .. py:attribute:: traces
153 Extracted waveforms for this time window.
154 '''
156 def __init__(self, tmin, tmax, i, n, igroup, ngroups, traces):
157 self.tmin = tmin
158 self.tmax = tmax
159 self.i = i
160 self.n = n
161 self.igroup = igroup
162 self.ngroups = ngroups
163 self.traces = traces
166class Squirrel(Selection):
167 '''
168 Prompt, lazy, indexing, caching, dynamic seismological dataset access.
170 :param env:
171 Squirrel environment instance or directory path to use as starting
172 point for its detection. By default, the current directory is used as
173 starting point. When searching for a usable environment the directory
174 ``'.squirrel'`` or ``'squirrel'`` in the current (or starting point)
175 directory is used if it exists, otherwise the parent directories are
176 search upwards for the existence of such a directory. If no such
177 directory is found, the user's global Squirrel environment
178 ``'$HOME/.pyrocko/squirrel'`` is used.
179 :type env:
180 :py:class:`~pyrocko.squirrel.environment.Environment` or
181 :py:class:`str`
183 :param database:
184 Database instance or path to database. By default the
185 database found in the detected Squirrel environment is used.
186 :type database:
187 :py:class:`~pyrocko.squirrel.database.Database` or :py:class:`str`
189 :param cache_path:
190 Directory path to use for data caching. By default, the ``'cache'``
191 directory in the detected Squirrel environment is used.
192 :type cache_path:
193 :py:class:`str`
195 :param persistent:
196 If given a name, create a persistent selection.
197 :type persistent:
198 :py:class:`str`
200 This is the central class of the Squirrel framework. It provides a unified
201 interface to query and access seismic waveforms, station meta-data and
202 event information from local file collections and remote data sources. For
203 prompt responses, a profound database setup is used under the hood. To
204 speed up assemblage of ad-hoc data selections, files are indexed on first
205 use and the extracted meta-data is remembered in the database for
206 subsequent accesses. Bulk data is lazily loaded from disk and remote
207 sources, just when requested. Once loaded, data is cached in memory to
208 expedite typical access patterns. Files and data sources can be dynamically
209 added to and removed from the Squirrel selection at runtime.
211 Queries are restricted to the contents of the files currently added to the
212 Squirrel selection (usually a subset of the file meta-information
213 collection in the database). This list of files is referred to here as the
214 "selection". By default, temporary tables are created in the attached
215 database to hold the names of the files in the selection as well as various
216 indices and counters. These tables are only visible inside the application
217 which created them and are deleted when the database connection is closed
218 or the application exits. To create a selection which is not deleted at
219 exit, supply a name to the ``persistent`` argument of the Squirrel
220 constructor. Persistent selections are shared among applications using the
221 same database.
223 **Method summary**
225 Some of the methods are implemented in :py:class:`Squirrel`'s base class
226 :py:class:`~pyrocko.squirrel.selection.Selection`.
228 .. autosummary::
230 ~Squirrel.add
231 ~Squirrel.add_source
232 ~Squirrel.add_fdsn
233 ~Squirrel.add_catalog
234 ~Squirrel.add_dataset
235 ~Squirrel.add_virtual
236 ~Squirrel.update
237 ~Squirrel.update_waveform_promises
238 ~Squirrel.advance_accessor
239 ~Squirrel.clear_accessor
240 ~Squirrel.reload
241 ~pyrocko.squirrel.selection.Selection.iter_paths
242 ~Squirrel.iter_nuts
243 ~Squirrel.iter_kinds
244 ~Squirrel.iter_deltats
245 ~Squirrel.iter_codes
246 ~pyrocko.squirrel.selection.Selection.get_paths
247 ~Squirrel.get_nuts
248 ~Squirrel.get_kinds
249 ~Squirrel.get_deltats
250 ~Squirrel.get_codes
251 ~Squirrel.get_counts
252 ~Squirrel.get_time_span
253 ~Squirrel.get_deltat_span
254 ~Squirrel.get_nfiles
255 ~Squirrel.get_nnuts
256 ~Squirrel.get_total_size
257 ~Squirrel.get_stats
258 ~Squirrel.get_content
259 ~Squirrel.get_stations
260 ~Squirrel.get_channels
261 ~Squirrel.get_responses
262 ~Squirrel.get_events
263 ~Squirrel.get_waveform_nuts
264 ~Squirrel.get_waveforms
265 ~Squirrel.chopper_waveforms
266 ~Squirrel.get_coverage
267 ~Squirrel.pile
268 ~Squirrel.snuffle
269 ~Squirrel.glob_codes
270 ~pyrocko.squirrel.selection.Selection.get_database
271 ~Squirrel.print_tables
272 '''
274 def __init__(
275 self, env=None, database=None, cache_path=None, persistent=None):
277 if not isinstance(env, environment.Environment):
278 env = environment.get_environment(env)
280 if database is None:
281 database = env.expand_path(env.database_path)
283 if cache_path is None:
284 cache_path = env.expand_path(env.cache_path)
286 if persistent is None:
287 persistent = env.persistent
289 Selection.__init__(
290 self, database=database, persistent=persistent)
292 self.get_database().set_basepath(os.path.dirname(env.get_basepath()))
294 self._content_caches = {
295 'waveform': cache.ContentCache(),
296 'default': cache.ContentCache()}
298 self._cache_path = cache_path
300 self._sources = []
301 self._operators = []
302 self._operator_registry = {}
304 self._pending_orders = []
306 self._pile = None
307 self._n_choppers_active = 0
309 self._names.update({
310 'nuts': self.name + '_nuts',
311 'kind_codes_count': self.name + '_kind_codes_count',
312 'coverage': self.name + '_coverage'})
314 with self.transaction('create tables') as cursor:
315 self._create_tables_squirrel(cursor)
317 def _create_tables_squirrel(self, cursor):
319 cursor.execute(self._register_table(self._sql(
320 '''
321 CREATE TABLE IF NOT EXISTS %(db)s.%(nuts)s (
322 nut_id integer PRIMARY KEY,
323 file_id integer,
324 file_segment integer,
325 file_element integer,
326 kind_id integer,
327 kind_codes_id integer,
328 tmin_seconds integer,
329 tmin_offset integer,
330 tmax_seconds integer,
331 tmax_offset integer,
332 kscale integer)
333 ''')))
335 cursor.execute(self._register_table(self._sql(
336 '''
337 CREATE TABLE IF NOT EXISTS %(db)s.%(kind_codes_count)s (
338 kind_codes_id integer PRIMARY KEY,
339 count integer)
340 ''')))
342 cursor.execute(self._sql(
343 '''
344 CREATE UNIQUE INDEX IF NOT EXISTS %(db)s.%(nuts)s_file_element
345 ON %(nuts)s (file_id, file_segment, file_element)
346 '''))
348 cursor.execute(self._sql(
349 '''
350 CREATE INDEX IF NOT EXISTS %(db)s.%(nuts)s_index_file_id
351 ON %(nuts)s (file_id)
352 '''))
354 cursor.execute(self._sql(
355 '''
356 CREATE INDEX IF NOT EXISTS %(db)s.%(nuts)s_index_tmin_seconds
357 ON %(nuts)s (kind_id, tmin_seconds)
358 '''))
360 cursor.execute(self._sql(
361 '''
362 CREATE INDEX IF NOT EXISTS %(db)s.%(nuts)s_index_tmax_seconds
363 ON %(nuts)s (kind_id, tmax_seconds)
364 '''))
366 cursor.execute(self._sql(
367 '''
368 CREATE INDEX IF NOT EXISTS %(db)s.%(nuts)s_index_kscale
369 ON %(nuts)s (kind_id, kscale, tmin_seconds)
370 '''))
372 cursor.execute(self._sql(
373 '''
374 CREATE TRIGGER IF NOT EXISTS %(db)s.%(nuts)s_delete_nuts
375 BEFORE DELETE ON main.files FOR EACH ROW
376 BEGIN
377 DELETE FROM %(nuts)s WHERE file_id == old.file_id;
378 END
379 '''))
381 # trigger only on size to make silent update of mtime possible
382 cursor.execute(self._sql(
383 '''
384 CREATE TRIGGER IF NOT EXISTS %(db)s.%(nuts)s_delete_nuts2
385 BEFORE UPDATE OF size ON main.files FOR EACH ROW
386 BEGIN
387 DELETE FROM %(nuts)s WHERE file_id == old.file_id;
388 END
389 '''))
391 cursor.execute(self._sql(
392 '''
393 CREATE TRIGGER IF NOT EXISTS
394 %(db)s.%(file_states)s_delete_files
395 BEFORE DELETE ON %(db)s.%(file_states)s FOR EACH ROW
396 BEGIN
397 DELETE FROM %(nuts)s WHERE file_id == old.file_id;
398 END
399 '''))
401 cursor.execute(self._sql(
402 '''
403 CREATE TRIGGER IF NOT EXISTS %(db)s.%(nuts)s_inc_kind_codes
404 BEFORE INSERT ON %(nuts)s FOR EACH ROW
405 BEGIN
406 INSERT OR IGNORE INTO %(kind_codes_count)s VALUES
407 (new.kind_codes_id, 0);
408 UPDATE %(kind_codes_count)s
409 SET count = count + 1
410 WHERE new.kind_codes_id
411 == %(kind_codes_count)s.kind_codes_id;
412 END
413 '''))
415 cursor.execute(self._sql(
416 '''
417 CREATE TRIGGER IF NOT EXISTS %(db)s.%(nuts)s_dec_kind_codes
418 BEFORE DELETE ON %(nuts)s FOR EACH ROW
419 BEGIN
420 UPDATE %(kind_codes_count)s
421 SET count = count - 1
422 WHERE old.kind_codes_id
423 == %(kind_codes_count)s.kind_codes_id;
424 END
425 '''))
427 cursor.execute(self._register_table(self._sql(
428 '''
429 CREATE TABLE IF NOT EXISTS %(db)s.%(coverage)s (
430 kind_codes_id integer,
431 time_seconds integer,
432 time_offset integer,
433 step integer)
434 ''')))
436 cursor.execute(self._sql(
437 '''
438 CREATE UNIQUE INDEX IF NOT EXISTS %(db)s.%(coverage)s_time
439 ON %(coverage)s (kind_codes_id, time_seconds, time_offset)
440 '''))
442 cursor.execute(self._sql(
443 '''
444 CREATE TRIGGER IF NOT EXISTS %(db)s.%(nuts)s_add_coverage
445 AFTER INSERT ON %(nuts)s FOR EACH ROW
446 BEGIN
447 INSERT OR IGNORE INTO %(coverage)s VALUES
448 (new.kind_codes_id, new.tmin_seconds, new.tmin_offset, 0)
449 ;
450 UPDATE %(coverage)s
451 SET step = step + 1
452 WHERE new.kind_codes_id == %(coverage)s.kind_codes_id
453 AND new.tmin_seconds == %(coverage)s.time_seconds
454 AND new.tmin_offset == %(coverage)s.time_offset
455 ;
456 INSERT OR IGNORE INTO %(coverage)s VALUES
457 (new.kind_codes_id, new.tmax_seconds, new.tmax_offset, 0)
458 ;
459 UPDATE %(coverage)s
460 SET step = step - 1
461 WHERE new.kind_codes_id == %(coverage)s.kind_codes_id
462 AND new.tmax_seconds == %(coverage)s.time_seconds
463 AND new.tmax_offset == %(coverage)s.time_offset
464 ;
465 DELETE FROM %(coverage)s
466 WHERE new.kind_codes_id == %(coverage)s.kind_codes_id
467 AND new.tmin_seconds == %(coverage)s.time_seconds
468 AND new.tmin_offset == %(coverage)s.time_offset
469 AND step == 0
470 ;
471 DELETE FROM %(coverage)s
472 WHERE new.kind_codes_id == %(coverage)s.kind_codes_id
473 AND new.tmax_seconds == %(coverage)s.time_seconds
474 AND new.tmax_offset == %(coverage)s.time_offset
475 AND step == 0
476 ;
477 END
478 '''))
480 cursor.execute(self._sql(
481 '''
482 CREATE TRIGGER IF NOT EXISTS %(db)s.%(nuts)s_remove_coverage
483 BEFORE DELETE ON %(nuts)s FOR EACH ROW
484 BEGIN
485 INSERT OR IGNORE INTO %(coverage)s VALUES
486 (old.kind_codes_id, old.tmin_seconds, old.tmin_offset, 0)
487 ;
488 UPDATE %(coverage)s
489 SET step = step - 1
490 WHERE old.kind_codes_id == %(coverage)s.kind_codes_id
491 AND old.tmin_seconds == %(coverage)s.time_seconds
492 AND old.tmin_offset == %(coverage)s.time_offset
493 ;
494 INSERT OR IGNORE INTO %(coverage)s VALUES
495 (old.kind_codes_id, old.tmax_seconds, old.tmax_offset, 0)
496 ;
497 UPDATE %(coverage)s
498 SET step = step + 1
499 WHERE old.kind_codes_id == %(coverage)s.kind_codes_id
500 AND old.tmax_seconds == %(coverage)s.time_seconds
501 AND old.tmax_offset == %(coverage)s.time_offset
502 ;
503 DELETE FROM %(coverage)s
504 WHERE old.kind_codes_id == %(coverage)s.kind_codes_id
505 AND old.tmin_seconds == %(coverage)s.time_seconds
506 AND old.tmin_offset == %(coverage)s.time_offset
507 AND step == 0
508 ;
509 DELETE FROM %(coverage)s
510 WHERE old.kind_codes_id == %(coverage)s.kind_codes_id
511 AND old.tmax_seconds == %(coverage)s.time_seconds
512 AND old.tmax_offset == %(coverage)s.time_offset
513 AND step == 0
514 ;
515 END
516 '''))
518 def _delete(self):
519 '''Delete database tables associated with this Squirrel.'''
521 with self.transaction('delete tables') as cursor:
522 for s in '''
523 DROP TRIGGER %(db)s.%(nuts)s_delete_nuts;
524 DROP TRIGGER %(db)s.%(nuts)s_delete_nuts2;
525 DROP TRIGGER %(db)s.%(file_states)s_delete_files;
526 DROP TRIGGER %(db)s.%(nuts)s_inc_kind_codes;
527 DROP TRIGGER %(db)s.%(nuts)s_dec_kind_codes;
528 DROP TABLE %(db)s.%(nuts)s;
529 DROP TABLE %(db)s.%(kind_codes_count)s;
530 DROP TRIGGER IF EXISTS %(db)s.%(nuts)s_add_coverage;
531 DROP TRIGGER IF EXISTS %(db)s.%(nuts)s_remove_coverage;
532 DROP TABLE IF EXISTS %(db)s.%(coverage)s;
533 '''.strip().splitlines():
535 cursor.execute(self._sql(s))
537 Selection._delete(self)
539 @filldocs
540 def add(self,
541 paths,
542 kinds=None,
543 format='detect',
544 include=None,
545 exclude=None,
546 check=True):
548 '''
549 Add files to the selection.
551 :param paths:
552 Iterator yielding paths to files or directories to be added to the
553 selection. Recurses into directories. If given a ``str``, it
554 is treated as a single path to be added.
555 :type paths:
556 :py:class:`list` of :py:class:`str`
558 :param kinds:
559 Content types to be made available through the Squirrel selection.
560 By default, all known content types are accepted.
561 :type kinds:
562 :py:class:`list` of :py:class:`str`
564 :param format:
565 File format identifier or ``'detect'`` to enable auto-detection
566 (available: %(file_formats)s).
567 :type format:
568 str
570 :param include:
571 If not ``None``, files are only included if their paths match the
572 given regular expression pattern.
573 :type format:
574 str
576 :param exclude:
577 If not ``None``, files are only included if their paths do not
578 match the given regular expression pattern.
579 :type format:
580 str
582 :param check:
583 If ``True``, all file modification times are checked to see if
584 cached information has to be updated (slow). If ``False``, only
585 previously unknown files are indexed and cached information is used
586 for known files, regardless of file state (fast, corrresponds to
587 Squirrel's ``--optimistic`` mode). File deletions will go
588 undetected in the latter case.
589 :type check:
590 bool
592 :Complexity:
593 O(log N)
594 '''
596 if isinstance(kinds, str):
597 kinds = (kinds,)
599 if isinstance(paths, str):
600 paths = [paths]
602 kind_mask = model.to_kind_mask(kinds)
604 with progress.view():
605 Selection.add(
606 self, util.iter_select_files(
607 paths,
608 show_progress=False,
609 include=include,
610 exclude=exclude,
611 pass_through=lambda path: path.startswith('virtual:')
612 ), kind_mask, format)
614 self._load(check)
615 self._update_nuts()
617 def reload(self):
618 '''
619 Check for modifications and reindex modified files.
621 Based on file modification times.
622 '''
624 self._set_file_states_force_check()
625 self._load(check=True)
626 self._update_nuts()
628 def add_virtual(self, nuts, virtual_paths=None):
629 '''
630 Add content which is not backed by files.
632 :param nuts:
633 Content pieces to be added.
634 :type nuts:
635 iterator yielding :py:class:`~pyrocko.squirrel.model.Nut` objects
637 :param virtual_paths:
638 List of virtual paths to prevent creating a temporary list of the
639 nuts while aggregating the file paths for the selection.
640 :type virtual_paths:
641 :py:class:`list` of :py:class:`str`
643 Stores to the main database and the selection.
644 '''
646 if isinstance(virtual_paths, str):
647 virtual_paths = [virtual_paths]
649 if virtual_paths is None:
650 if not isinstance(nuts, list):
651 nuts = list(nuts)
652 virtual_paths = set(nut.file_path for nut in nuts)
654 Selection.add(self, virtual_paths)
655 self.get_database().dig(nuts)
656 self._update_nuts()
658 def add_volatile(self, nuts):
659 if not isinstance(nuts, list):
660 nuts = list(nuts)
662 paths = list(set(nut.file_path for nut in nuts))
663 io.backends.virtual.add_nuts(nuts)
664 self.add_virtual(nuts, paths)
665 self._volatile_paths.extend(paths)
667 def add_volatile_waveforms(self, traces):
668 '''
669 Add in-memory waveforms which will be removed when the app closes.
670 '''
672 name = model.random_name()
674 path = 'virtual:volatile:%s' % name
676 nuts = []
677 for itr, tr in enumerate(traces):
678 assert tr.tmin <= tr.tmax
679 tmin_seconds, tmin_offset = model.tsplit(tr.tmin)
680 tmax_seconds, tmax_offset = model.tsplit(
681 tr.tmin + tr.data_len()*tr.deltat)
683 nuts.append(model.Nut(
684 file_path=path,
685 file_format='virtual',
686 file_segment=itr,
687 file_element=0,
688 file_mtime=0,
689 codes=tr.codes,
690 tmin_seconds=tmin_seconds,
691 tmin_offset=tmin_offset,
692 tmax_seconds=tmax_seconds,
693 tmax_offset=tmax_offset,
694 deltat=tr.deltat,
695 kind_id=to_kind_id('waveform'),
696 content=tr))
698 self.add_volatile(nuts)
699 return path
701 def _load(self, check):
702 for _ in io.iload(
703 self,
704 content=[],
705 skip_unchanged=True,
706 check=check):
707 pass
709 def _update_nuts(self, transaction=None):
710 transaction = transaction or self.transaction('update nuts')
711 with make_task('Aggregating selection') as task, \
712 transaction as cursor:
714 self._conn.set_progress_handler(task.update, 100000)
715 nrows = cursor.execute(self._sql(
716 '''
717 INSERT INTO %(db)s.%(nuts)s
718 SELECT NULL,
719 nuts.file_id, nuts.file_segment, nuts.file_element,
720 nuts.kind_id, nuts.kind_codes_id,
721 nuts.tmin_seconds, nuts.tmin_offset,
722 nuts.tmax_seconds, nuts.tmax_offset,
723 nuts.kscale
724 FROM %(db)s.%(file_states)s
725 INNER JOIN nuts
726 ON %(db)s.%(file_states)s.file_id == nuts.file_id
727 INNER JOIN kind_codes
728 ON nuts.kind_codes_id ==
729 kind_codes.kind_codes_id
730 WHERE %(db)s.%(file_states)s.file_state != 2
731 AND (((1 << kind_codes.kind_id)
732 & %(db)s.%(file_states)s.kind_mask) != 0)
733 ''')).rowcount
735 task.update(nrows)
736 self._set_file_states_known(transaction)
737 self._conn.set_progress_handler(None, 0)
739 def add_source(self, source, check=True):
740 '''
741 Add remote resource.
743 :param source:
744 Remote data access client instance.
745 :type source:
746 subclass of :py:class:`~pyrocko.squirrel.client.base.Source`
747 '''
749 self._sources.append(source)
750 source.setup(self, check=check)
752 def add_fdsn(self, *args, **kwargs):
753 '''
754 Add FDSN site for transparent remote data access.
756 Arguments are passed to
757 :py:class:`~pyrocko.squirrel.client.fdsn.FDSNSource`.
758 '''
760 self.add_source(fdsn.FDSNSource(*args, **kwargs))
762 def add_catalog(self, *args, **kwargs):
763 '''
764 Add online catalog for transparent event data access.
766 Arguments are passed to
767 :py:class:`~pyrocko.squirrel.client.catalog.CatalogSource`.
768 '''
770 self.add_source(catalog.CatalogSource(*args, **kwargs))
772 def add_dataset(self, ds, check=True):
773 '''
774 Read dataset description from file and add its contents.
776 :param ds:
777 Path to dataset description file or dataset description object
778 . See :py:mod:`~pyrocko.squirrel.dataset`.
779 :type ds:
780 :py:class:`str` or :py:class:`~pyrocko.squirrel.dataset.Dataset`
782 :param check:
783 If ``True``, all file modification times are checked to see if
784 cached information has to be updated (slow). If ``False``, only
785 previously unknown files are indexed and cached information is used
786 for known files, regardless of file state (fast, corrresponds to
787 Squirrel's ``--optimistic`` mode). File deletions will go
788 undetected in the latter case.
789 :type check:
790 bool
791 '''
792 if isinstance(ds, str):
793 ds = dataset.read_dataset(ds)
795 ds.setup(self, check=check)
797 def _get_selection_args(
798 self, kind_id,
799 obj=None, tmin=None, tmax=None, time=None, codes=None):
801 if codes is not None:
802 codes = codes_patterns_for_kind(kind_id, codes)
804 if time is not None:
805 tmin = time
806 tmax = time
808 if obj is not None:
809 tmin = tmin if tmin is not None else obj.tmin
810 tmax = tmax if tmax is not None else obj.tmax
811 codes = codes if codes is not None else codes_patterns_for_kind(
812 kind_id, obj.codes)
814 return tmin, tmax, codes
816 def _get_selection_args_str(self, *args, **kwargs):
818 tmin, tmax, codes = self._get_selection_args(*args, **kwargs)
819 return 'tmin: %s, tmax: %s, codes: %s' % (
820 util.time_to_str(tmin) if tmin is not None else 'none',
821 util.time_to_str(tmax) if tmax is not None else 'none',
822 ','.join(str(entry) for entry in codes))
824 def _selection_args_to_kwargs(
825 self, obj=None, tmin=None, tmax=None, time=None, codes=None):
827 return dict(obj=obj, tmin=tmin, tmax=tmax, time=time, codes=codes)
829 def _timerange_sql(self, tmin, tmax, kind, cond, args, naiv):
831 tmin_seconds, tmin_offset = model.tsplit(tmin)
832 tmax_seconds, tmax_offset = model.tsplit(tmax)
833 if naiv:
834 cond.append('%(db)s.%(nuts)s.tmin_seconds <= ?')
835 args.append(tmax_seconds)
836 else:
837 tscale_edges = model.tscale_edges
838 tmin_cond = []
839 for kscale in range(tscale_edges.size + 1):
840 if kscale != tscale_edges.size:
841 tscale = int(tscale_edges[kscale])
842 tmin_cond.append('''
843 (%(db)s.%(nuts)s.kind_id = ?
844 AND %(db)s.%(nuts)s.kscale == ?
845 AND %(db)s.%(nuts)s.tmin_seconds BETWEEN ? AND ?)
846 ''')
847 args.extend(
848 (to_kind_id(kind), kscale,
849 tmin_seconds - tscale - 1, tmax_seconds + 1))
851 else:
852 tmin_cond.append('''
853 (%(db)s.%(nuts)s.kind_id == ?
854 AND %(db)s.%(nuts)s.kscale == ?
855 AND %(db)s.%(nuts)s.tmin_seconds <= ?)
856 ''')
858 args.extend(
859 (to_kind_id(kind), kscale, tmax_seconds + 1))
860 if tmin_cond:
861 cond.append(' ( ' + ' OR '.join(tmin_cond) + ' ) ')
863 cond.append('%(db)s.%(nuts)s.tmax_seconds >= ?')
864 args.append(tmin_seconds)
866 def _codes_match_sql(self, kind_id, codes, cond, args):
867 pats = codes_patterns_for_kind(kind_id, codes)
868 if pats is None:
869 return
871 pats_exact = []
872 pats_nonexact = []
873 for pat in pats:
874 spat = pat.safe_str
875 (pats_exact if _is_exact(spat) else pats_nonexact).append(spat)
877 cond_exact = None
878 if pats_exact:
879 cond_exact = ' ( kind_codes.codes IN ( %s ) ) ' % ', '.join(
880 '?'*len(pats_exact))
882 args.extend(pats_exact)
884 cond_nonexact = None
885 if pats_nonexact:
886 cond_nonexact = ' ( %s ) ' % ' OR '.join(
887 ('kind_codes.codes GLOB ?',) * len(pats_nonexact))
889 args.extend(pats_nonexact)
891 if cond_exact and cond_nonexact:
892 cond.append(' ( %s OR %s ) ' % (cond_exact, cond_nonexact))
894 elif cond_exact:
895 cond.append(cond_exact)
897 elif cond_nonexact:
898 cond.append(cond_nonexact)
900 def iter_nuts(
901 self, kind=None, tmin=None, tmax=None, codes=None, naiv=False,
902 kind_codes_ids=None, path=None, limit=None):
904 '''
905 Iterate over content entities matching given constraints.
907 :param kind:
908 Content kind (or kinds) to extract.
909 :type kind:
910 :py:class:`str`, :py:class:`list` of :py:class:`str`
912 :param tmin:
913 Start time of query interval.
914 :type tmin:
915 timestamp
917 :param tmax:
918 End time of query interval.
919 :type tmax:
920 timestamp
922 :param codes:
923 List of code patterns to query.
924 :type codes:
925 :py:class:`list` of :py:class:`~pyrocko.squirrel.model.Codes`
926 objects appropriate for the queried content type, or anything which
927 can be converted to such objects.
929 :param naiv:
930 Bypass time span lookup through indices (slow, for testing).
931 :type naiv:
932 :py:class:`bool`
934 :param kind_codes_ids:
935 Kind-codes IDs of contents to be retrieved (internal use).
936 :type kind_codes_ids:
937 :py:class:`list` of :py:class:`int`
939 :yields:
940 :py:class:`~pyrocko.squirrel.model.Nut` objects representing the
941 intersecting content.
943 :complexity:
944 O(log N) for the time selection part due to heavy use of database
945 indices.
947 Query time span is treated as a half-open interval ``[tmin, tmax)``.
948 However, if ``tmin`` equals ``tmax``, the edge logics are modified to
949 closed-interval so that content intersecting with the time instant ``t
950 = tmin = tmax`` is returned (otherwise nothing would be returned as
951 ``[t, t)`` never matches anything).
953 Time spans of content entities to be matched are also treated as half
954 open intervals, e.g. content span ``[0, 1)`` is matched by query span
955 ``[0, 1)`` but not by ``[-1, 0)`` or ``[1, 2)``. Also here, logics are
956 modified to closed-interval when the content time span is an empty
957 interval, i.e. to indicate a time instant. E.g. time instant 0 is
958 matched by ``[0, 1)`` but not by ``[-1, 0)`` or ``[1, 2)``.
959 '''
961 if not isinstance(kind, str):
962 if kind is None:
963 kind = model.g_content_kinds
964 for kind_ in kind:
965 for nut in self.iter_nuts(kind_, tmin, tmax, codes):
966 yield nut
968 return
970 kind_id = to_kind_id(kind)
972 cond = []
973 args = []
974 if tmin is not None or tmax is not None:
975 assert kind is not None
976 if tmin is None:
977 tmin = self.get_time_span()[0]
978 if tmax is None:
979 tmax = self.get_time_span()[1] + 1.0
981 self._timerange_sql(tmin, tmax, kind, cond, args, naiv)
983 cond.append('kind_codes.kind_id == ?')
984 args.append(kind_id)
986 if codes is not None:
987 self._codes_match_sql(kind_id, codes, cond, args)
989 if kind_codes_ids is not None:
990 cond.append(
991 ' ( kind_codes.kind_codes_id IN ( %s ) ) ' % ', '.join(
992 '?'*len(kind_codes_ids)))
994 args.extend(kind_codes_ids)
996 db = self.get_database()
997 if path is not None:
998 cond.append('files.path == ?')
999 args.append(db.relpath(abspath(path)))
1001 sql = ('''
1002 SELECT
1003 files.path,
1004 files.format,
1005 files.mtime,
1006 files.size,
1007 %(db)s.%(nuts)s.file_segment,
1008 %(db)s.%(nuts)s.file_element,
1009 kind_codes.kind_id,
1010 kind_codes.codes,
1011 %(db)s.%(nuts)s.tmin_seconds,
1012 %(db)s.%(nuts)s.tmin_offset,
1013 %(db)s.%(nuts)s.tmax_seconds,
1014 %(db)s.%(nuts)s.tmax_offset,
1015 kind_codes.deltat
1016 FROM files
1017 INNER JOIN %(db)s.%(nuts)s
1018 ON files.file_id == %(db)s.%(nuts)s.file_id
1019 INNER JOIN kind_codes
1020 ON %(db)s.%(nuts)s.kind_codes_id == kind_codes.kind_codes_id
1021 ''')
1023 if cond:
1024 sql += ''' WHERE ''' + ' AND '.join(cond)
1026 if limit is not None:
1027 sql += ''' LIMIT %i''' % limit
1029 sql = self._sql(sql)
1030 if tmin is None and tmax is None:
1031 for row in self._conn.execute(sql, args):
1032 row = (db.abspath(row[0]),) + row[1:]
1033 nut = model.Nut(values_nocheck=row)
1034 yield nut
1035 else:
1036 assert tmin is not None and tmax is not None
1037 if tmin == tmax:
1038 for row in self._conn.execute(sql, args):
1039 row = (db.abspath(row[0]),) + row[1:]
1040 nut = model.Nut(values_nocheck=row)
1041 if (nut.tmin <= tmin < nut.tmax) \
1042 or (nut.tmin == nut.tmax and tmin == nut.tmin):
1044 yield nut
1045 else:
1046 for row in self._conn.execute(sql, args):
1047 row = (db.abspath(row[0]),) + row[1:]
1048 nut = model.Nut(values_nocheck=row)
1049 if (tmin < nut.tmax and nut.tmin < tmax) \
1050 or (nut.tmin == nut.tmax
1051 and tmin <= nut.tmin < tmax):
1053 yield nut
1055 def get_nuts(self, *args, **kwargs):
1056 '''
1057 Get content entities matching given constraints.
1059 Like :py:meth:`iter_nuts` but returns results as a list.
1060 '''
1062 return list(self.iter_nuts(*args, **kwargs))
1064 def _split_nuts(
1065 self, kind, tmin=None, tmax=None, codes=None, path=None):
1067 kind_id = to_kind_id(kind)
1068 tmin_seconds, tmin_offset = model.tsplit(tmin)
1069 tmax_seconds, tmax_offset = model.tsplit(tmax)
1071 names_main_nuts = dict(self._names)
1072 names_main_nuts.update(db='main', nuts='nuts')
1074 db = self.get_database()
1076 def main_nuts(s):
1077 return s % names_main_nuts
1079 with self.transaction('split nuts') as cursor:
1080 # modify selection and main
1081 for sql_subst in [
1082 self._sql, main_nuts]:
1084 cond = []
1085 args = []
1087 self._timerange_sql(tmin, tmax, kind, cond, args, False)
1089 if codes is not None:
1090 self._codes_match_sql(kind_id, codes, cond, args)
1092 if path is not None:
1093 cond.append('files.path == ?')
1094 args.append(db.relpath(abspath(path)))
1096 sql = sql_subst('''
1097 SELECT
1098 %(db)s.%(nuts)s.nut_id,
1099 %(db)s.%(nuts)s.tmin_seconds,
1100 %(db)s.%(nuts)s.tmin_offset,
1101 %(db)s.%(nuts)s.tmax_seconds,
1102 %(db)s.%(nuts)s.tmax_offset,
1103 kind_codes.deltat
1104 FROM files
1105 INNER JOIN %(db)s.%(nuts)s
1106 ON files.file_id == %(db)s.%(nuts)s.file_id
1107 INNER JOIN kind_codes
1108 ON %(db)s.%(nuts)s.kind_codes_id == kind_codes.kind_codes_id
1109 WHERE ''' + ' AND '.join(cond)) # noqa
1111 insert = []
1112 delete = []
1113 for row in cursor.execute(sql, args):
1114 nut_id, nut_tmin_seconds, nut_tmin_offset, \
1115 nut_tmax_seconds, nut_tmax_offset, nut_deltat = row
1117 nut_tmin = model.tjoin(
1118 nut_tmin_seconds, nut_tmin_offset)
1119 nut_tmax = model.tjoin(
1120 nut_tmax_seconds, nut_tmax_offset)
1122 if nut_tmin < tmax and tmin < nut_tmax:
1123 if nut_tmin < tmin:
1124 insert.append((
1125 nut_tmin_seconds, nut_tmin_offset,
1126 tmin_seconds, tmin_offset,
1127 model.tscale_to_kscale(
1128 tmin_seconds - nut_tmin_seconds),
1129 nut_id))
1131 if tmax < nut_tmax:
1132 insert.append((
1133 tmax_seconds, tmax_offset,
1134 nut_tmax_seconds, nut_tmax_offset,
1135 model.tscale_to_kscale(
1136 nut_tmax_seconds - tmax_seconds),
1137 nut_id))
1139 delete.append((nut_id,))
1141 sql_add = '''
1142 INSERT INTO %(db)s.%(nuts)s (
1143 file_id, file_segment, file_element, kind_id,
1144 kind_codes_id, tmin_seconds, tmin_offset,
1145 tmax_seconds, tmax_offset, kscale )
1146 SELECT
1147 file_id, file_segment, file_element,
1148 kind_id, kind_codes_id, ?, ?, ?, ?, ?
1149 FROM %(db)s.%(nuts)s
1150 WHERE nut_id == ?
1151 '''
1152 cursor.executemany(sql_subst(sql_add), insert)
1154 sql_delete = '''
1155 DELETE FROM %(db)s.%(nuts)s WHERE nut_id == ?
1156 '''
1157 cursor.executemany(sql_subst(sql_delete), delete)
1159 def get_time_span(self, kinds=None):
1160 '''
1161 Get time interval over all content in selection.
1163 :param kinds:
1164 If not ``None``, restrict query to given content kinds.
1165 :type kind:
1166 list of str
1168 :complexity:
1169 O(1), independent of the number of nuts.
1171 :returns:
1172 ``(tmin, tmax)``, combined time interval of queried content kinds.
1173 '''
1175 sql_min = self._sql('''
1176 SELECT MIN(tmin_seconds), MIN(tmin_offset)
1177 FROM %(db)s.%(nuts)s
1178 WHERE kind_id == ?
1179 AND tmin_seconds == (
1180 SELECT MIN(tmin_seconds)
1181 FROM %(db)s.%(nuts)s
1182 WHERE kind_id == ?)
1183 ''')
1185 sql_max = self._sql('''
1186 SELECT MAX(tmax_seconds), MAX(tmax_offset)
1187 FROM %(db)s.%(nuts)s
1188 WHERE kind_id == ?
1189 AND tmax_seconds == (
1190 SELECT MAX(tmax_seconds)
1191 FROM %(db)s.%(nuts)s
1192 WHERE kind_id == ?)
1193 ''')
1195 gtmin = None
1196 gtmax = None
1198 if isinstance(kinds, str):
1199 kinds = [kinds]
1201 if kinds is None:
1202 kind_ids = model.g_content_kind_ids
1203 else:
1204 kind_ids = model.to_kind_ids(kinds)
1206 for kind_id in kind_ids:
1207 for tmin_seconds, tmin_offset in self._conn.execute(
1208 sql_min, (kind_id, kind_id)):
1209 tmin = model.tjoin(tmin_seconds, tmin_offset)
1210 if tmin is not None and (gtmin is None or tmin < gtmin):
1211 gtmin = tmin
1213 for (tmax_seconds, tmax_offset) in self._conn.execute(
1214 sql_max, (kind_id, kind_id)):
1215 tmax = model.tjoin(tmax_seconds, tmax_offset)
1216 if tmax is not None and (gtmax is None or tmax > gtmax):
1217 gtmax = tmax
1219 return gtmin, gtmax
1221 def has(self, kinds):
1222 '''
1223 Check availability of given content kinds.
1225 :param kinds:
1226 Content kinds to query.
1227 :type kind:
1228 list of str
1230 :returns:
1231 ``True`` if any of the queried content kinds is available
1232 in the selection.
1233 '''
1234 self_tmin, self_tmax = self.get_time_span(kinds)
1236 return None not in (self_tmin, self_tmax)
1238 def get_deltat_span(self, kind):
1239 '''
1240 Get min and max sampling interval of all content of given kind.
1242 :param kind:
1243 Content kind
1244 :type kind:
1245 str
1247 :returns: ``(deltat_min, deltat_max)``
1248 '''
1250 deltats = [
1251 deltat for deltat in self.get_deltats(kind)
1252 if deltat is not None]
1254 if deltats:
1255 return min(deltats), max(deltats)
1256 else:
1257 return None, None
1259 def iter_kinds(self, codes=None):
1260 '''
1261 Iterate over content types available in selection.
1263 :param codes:
1264 If given, get kinds only for selected codes identifier.
1265 Only a single identifier may be given here and no pattern matching
1266 is done, currently.
1267 :type codes:
1268 :py:class:`~pyrocko.squirrel.model.Codes`
1270 :yields:
1271 Available content kinds as :py:class:`str`.
1273 :complexity:
1274 O(1), independent of number of nuts.
1275 '''
1277 return self._database._iter_kinds(
1278 codes=codes,
1279 kind_codes_count='%(db)s.%(kind_codes_count)s' % self._names)
1281 def iter_deltats(self, kind=None):
1282 '''
1283 Iterate over sampling intervals available in selection.
1285 :param kind:
1286 If given, get sampling intervals only for a given content type.
1287 :type kind:
1288 str
1290 :yields:
1291 :py:class:`float` values.
1293 :complexity:
1294 O(1), independent of number of nuts.
1295 '''
1296 return self._database._iter_deltats(
1297 kind=kind,
1298 kind_codes_count='%(db)s.%(kind_codes_count)s' % self._names)
1300 def iter_codes(self, kind=None):
1301 '''
1302 Iterate over content identifier code sequences available in selection.
1304 :param kind:
1305 If given, get codes only for a given content type.
1306 :type kind:
1307 str
1309 :yields:
1310 :py:class:`tuple` of :py:class:`str`
1312 :complexity:
1313 O(1), independent of number of nuts.
1314 '''
1315 return self._database._iter_codes(
1316 kind=kind,
1317 kind_codes_count='%(db)s.%(kind_codes_count)s' % self._names)
1319 def _iter_codes_info(self, kind=None, codes=None):
1320 '''
1321 Iterate over number of occurrences of any (kind, codes) combination.
1323 :param kind:
1324 If given, get counts only for selected content type.
1325 :type kind:
1326 str
1328 :yields:
1329 Tuples of the form ``(kind, codes, deltat, kind_codes_id, count)``.
1331 :complexity:
1332 O(1), independent of number of nuts.
1333 '''
1334 return self._database._iter_codes_info(
1335 kind=kind,
1336 codes=codes,
1337 kind_codes_count='%(db)s.%(kind_codes_count)s' % self._names)
1339 def get_kinds(self, codes=None):
1340 '''
1341 Get content types available in selection.
1343 :param codes:
1344 If given, get kinds only for selected codes identifier.
1345 Only a single identifier may be given here and no pattern matching
1346 is done, currently.
1347 :type codes:
1348 :py:class:`~pyrocko.squirrel.model.Codes`
1350 :returns:
1351 Sorted list of available content types.
1352 :rtype:
1353 py:class:`list` of :py:class:`str`
1355 :complexity:
1356 O(1), independent of number of nuts.
1358 '''
1359 return sorted(list(self.iter_kinds(codes=codes)))
1361 def get_deltats(self, kind=None):
1362 '''
1363 Get sampling intervals available in selection.
1365 :param kind:
1366 If given, get sampling intervals only for selected content type.
1367 :type kind:
1368 str
1370 :complexity:
1371 O(1), independent of number of nuts.
1373 :returns: Sorted list of available sampling intervals.
1374 '''
1375 return sorted(list(self.iter_deltats(kind=kind)))
1377 def get_codes(self, kind=None):
1378 '''
1379 Get identifier code sequences available in selection.
1381 :param kind:
1382 If given, get codes only for selected content type.
1383 :type kind:
1384 str
1386 :complexity:
1387 O(1), independent of number of nuts.
1389 :returns: Sorted list of available codes as tuples of strings.
1390 '''
1391 return sorted(list(self.iter_codes(kind=kind)))
1393 def get_counts(self, kind=None):
1394 '''
1395 Get number of occurrences of any (kind, codes) combination.
1397 :param kind:
1398 If given, get codes only for selected content type.
1399 :type kind:
1400 str
1402 :complexity:
1403 O(1), independent of number of nuts.
1405 :returns: ``dict`` with ``counts[kind][codes]`` or ``counts[codes]``
1406 if kind is not ``None``
1407 '''
1408 d = {}
1409 for kind_id, codes, _, _, count in self._iter_codes_info(kind=kind):
1410 if kind_id not in d:
1411 v = d[kind_id] = {}
1412 else:
1413 v = d[kind_id]
1415 if codes not in v:
1416 v[codes] = 0
1418 v[codes] += count
1420 if kind is not None:
1421 return d[to_kind_id(kind)]
1422 else:
1423 return dict((to_kind(kind_id), v) for (kind_id, v) in d.items())
1425 def glob_codes(self, kind, codes):
1426 '''
1427 Find codes matching given patterns.
1429 :param kind:
1430 Content kind to be queried.
1431 :type kind:
1432 str
1434 :param codes:
1435 List of code patterns to query.
1436 :type codes:
1437 :py:class:`list` of :py:class:`~pyrocko.squirrel.model.Codes`
1438 objects appropriate for the queried content type, or anything which
1439 can be converted to such objects.
1441 :returns:
1442 List of matches of the form ``[kind_codes_id, codes, deltat]``.
1443 '''
1445 kind_id = to_kind_id(kind)
1446 args = [kind_id]
1447 pats = codes_patterns_for_kind(kind_id, codes)
1449 if pats:
1450 codes_cond = 'AND ( %s ) ' % ' OR '.join(
1451 ('kind_codes.codes GLOB ?',) * len(pats))
1453 args.extend(pat.safe_str for pat in pats)
1454 else:
1455 codes_cond = ''
1457 sql = self._sql('''
1458 SELECT kind_codes_id, codes, deltat FROM kind_codes
1459 WHERE
1460 kind_id == ? ''' + codes_cond)
1462 return list(map(list, self._conn.execute(sql, args)))
1464 def update(self, constraint=None, **kwargs):
1465 '''
1466 Update or partially update channel and event inventories.
1468 :param constraint:
1469 Selection of times or areas to be brought up to date.
1470 :type constraint:
1471 :py:class:`~pyrocko.squirrel.client.base.Constraint`
1473 :param \\*\\*kwargs:
1474 Shortcut for setting ``constraint=Constraint(**kwargs)``.
1476 This function triggers all attached remote sources, to check for
1477 updates in the meta-data. The sources will only submit queries when
1478 their expiration date has passed, or if the selection spans into
1479 previously unseen times or areas.
1480 '''
1482 if constraint is None:
1483 constraint = client.Constraint(**kwargs)
1485 for source in self._sources:
1486 source.update_channel_inventory(self, constraint)
1487 source.update_event_inventory(self, constraint)
1489 def update_waveform_promises(self, constraint=None, **kwargs):
1490 '''
1491 Permit downloading of remote waveforms.
1493 :param constraint:
1494 Remote waveforms compatible with the given constraint are enabled
1495 for download.
1496 :type constraint:
1497 :py:class:`~pyrocko.squirrel.client.base.Constraint`
1499 :param \\*\\*kwargs:
1500 Shortcut for setting ``constraint=Constraint(**kwargs)``.
1502 Calling this method permits Squirrel to download waveforms from remote
1503 sources when processing subsequent waveform requests. This works by
1504 inserting so called waveform promises into the database. It will look
1505 into the available channels for each remote source and create a promise
1506 for each channel compatible with the given constraint. If the promise
1507 then matches in a waveform request, Squirrel tries to download the
1508 waveform. If the download is successful, the downloaded waveform is
1509 added to the Squirrel and the promise is deleted. If the download
1510 fails, the promise is kept if the reason of failure looks like being
1511 temporary, e.g. because of a network failure. If the cause of failure
1512 however seems to be permanent, the promise is deleted so that no
1513 further attempts are made to download a waveform which might not be
1514 available from that server at all. To force re-scheduling after a
1515 permanent failure, call :py:meth:`update_waveform_promises`
1516 yet another time.
1517 '''
1519 if constraint is None:
1520 constraint = client.Constraint(**kwargs)
1522 for source in self._sources:
1523 source.update_waveform_promises(self, constraint)
1525 def remove_waveform_promises(self, from_database='selection'):
1526 '''
1527 Remove waveform promises from live selection or global database.
1529 Calling this function removes all waveform promises provided by the
1530 attached sources.
1532 :param from_database:
1533 Remove from live selection ``'selection'`` or global database
1534 ``'global'``.
1535 '''
1536 for source in self._sources:
1537 source.remove_waveform_promises(self, from_database=from_database)
1539 def update_responses(self, constraint=None, **kwargs):
1540 if constraint is None:
1541 constraint = client.Constraint(**kwargs)
1543 for source in self._sources:
1544 source.update_response_inventory(self, constraint)
1546 def get_nfiles(self):
1547 '''
1548 Get number of files in selection.
1549 '''
1551 sql = self._sql('''SELECT COUNT(*) FROM %(db)s.%(file_states)s''')
1552 for row in self._conn.execute(sql):
1553 return row[0]
1555 def get_nnuts(self):
1556 '''
1557 Get number of nuts in selection.
1558 '''
1560 sql = self._sql('''SELECT COUNT(*) FROM %(db)s.%(nuts)s''')
1561 for row in self._conn.execute(sql):
1562 return row[0]
1564 def get_total_size(self):
1565 '''
1566 Get aggregated file size available in selection.
1567 '''
1569 sql = self._sql('''
1570 SELECT SUM(files.size) FROM %(db)s.%(file_states)s
1571 INNER JOIN files
1572 ON %(db)s.%(file_states)s.file_id = files.file_id
1573 ''')
1575 for row in self._conn.execute(sql):
1576 return row[0] or 0
1578 def get_stats(self):
1579 '''
1580 Get statistics on contents available through this selection.
1581 '''
1583 kinds = self.get_kinds()
1584 time_spans = {}
1585 for kind in kinds:
1586 time_spans[kind] = self.get_time_span([kind])
1588 return SquirrelStats(
1589 nfiles=self.get_nfiles(),
1590 nnuts=self.get_nnuts(),
1591 kinds=kinds,
1592 codes=self.get_codes(),
1593 total_size=self.get_total_size(),
1594 counts=self.get_counts(),
1595 time_spans=time_spans,
1596 sources=[s.describe() for s in self._sources],
1597 operators=[op.describe() for op in self._operators])
1599 @filldocs
1600 def check(
1601 self, obj=None, tmin=None, tmax=None, time=None, codes=None,
1602 ignore=[]):
1603 '''
1604 Check for common data/metadata problems.
1606 %(query_args)s
1608 :param ignore:
1609 Problem types to be ignored.
1610 :type ignore:
1611 :class:`list` of :class:`str`
1612 (:py:class:`~pyrocko.squirrel.check.SquirrelCheckProblemType`)
1614 :returns:
1615 :py:class:`~pyrocko.squirrel.check.SquirrelCheck` object
1616 containing the results of the check.
1618 See :py:func:`~pyrocko.squirrel.check.do_check`.
1619 '''
1621 from .check import do_check
1622 tmin, tmax, codes = self._get_selection_args(
1623 CHANNEL, obj, tmin, tmax, time, codes)
1625 return do_check(self, tmin=tmin, tmax=tmax, codes=codes, ignore=ignore)
1627 def get_content(
1628 self,
1629 nut,
1630 cache_id='default',
1631 accessor_id='default',
1632 show_progress=False,
1633 model='squirrel'):
1635 '''
1636 Get and possibly load full content for a given index entry from file.
1638 Loads the actual content objects (channel, station, waveform, ...) from
1639 file. For efficiency, sibling content (all stuff in the same file
1640 segment) will also be loaded as a side effect. The loaded contents are
1641 cached in the Squirrel object.
1642 '''
1644 content_cache = self._content_caches[cache_id]
1645 if not content_cache.has(nut):
1647 for nut_loaded in io.iload(
1648 nut.file_path,
1649 segment=nut.file_segment,
1650 format=nut.file_format,
1651 database=self._database,
1652 update_selection=self,
1653 show_progress=show_progress):
1655 content_cache.put(nut_loaded)
1657 try:
1658 return content_cache.get(nut, accessor_id, model)
1660 except KeyError:
1661 raise error.NotAvailable(
1662 'Unable to retrieve content: %s, %s, %s, %s' % nut.key)
1664 def advance_accessor(self, accessor_id='default', cache_id=None):
1665 '''
1666 Notify memory caches about consumer moving to a new data batch.
1668 :param accessor_id:
1669 Name of accessing consumer to be advanced.
1670 :type accessor_id:
1671 str
1673 :param cache_id:
1674 Name of cache to for which the accessor should be advanced. By
1675 default the named accessor is advanced in all registered caches.
1676 By default, two caches named ``'default'`` and ``'waveform'`` are
1677 available.
1678 :type cache_id:
1679 str
1681 See :py:class:`~pyrocko.squirrel.cache.ContentCache` for details on how
1682 Squirrel's memory caching works and can be tuned. Default behaviour is
1683 to release data when it has not been used in the latest data
1684 window/batch. If the accessor is never advanced, data is cached
1685 indefinitely - which is often desired e.g. for station meta-data.
1686 Methods for consecutive data traversal, like
1687 :py:meth:`chopper_waveforms` automatically advance and clear
1688 their accessor.
1689 '''
1690 for cache_ in (
1691 self._content_caches.keys()
1692 if cache_id is None
1693 else [cache_id]):
1695 self._content_caches[cache_].advance_accessor(accessor_id)
1697 def clear_accessor(self, accessor_id, cache_id=None):
1698 '''
1699 Notify memory caches about a consumer having finished.
1701 :param accessor_id:
1702 Name of accessor to be cleared.
1703 :type accessor_id:
1704 str
1706 :param cache_id:
1707 Name of cache for which the accessor should be cleared. By default
1708 the named accessor is cleared from all registered caches. By
1709 default, two caches named ``'default'`` and ``'waveform'`` are
1710 available.
1711 :type cache_id:
1712 str
1714 Calling this method clears all references to cache entries held by the
1715 named accessor. Cache entries are then freed if not referenced by any
1716 other accessor.
1717 '''
1719 for cache_ in (
1720 self._content_caches.keys()
1721 if cache_id is None
1722 else [cache_id]):
1724 self._content_caches[cache_].clear_accessor(accessor_id)
1726 def get_cache_stats(self, cache_id):
1727 return self._content_caches[cache_id].get_stats()
1729 @filldocs
1730 def get_stations(
1731 self, obj=None, tmin=None, tmax=None, time=None, codes=None,
1732 model='squirrel'):
1734 '''
1735 Get stations matching given constraints.
1737 %(query_args)s
1739 :param model:
1740 Select object model for returned values: ``'squirrel'`` to get
1741 Squirrel station objects or ``'pyrocko'`` to get Pyrocko station
1742 objects with channel information attached.
1743 :type model:
1744 str
1746 :returns:
1747 List of :py:class:`pyrocko.squirrel.Station
1748 <pyrocko.squirrel.model.Station>` objects by default or list of
1749 :py:class:`pyrocko.model.Station <pyrocko.model.station.Station>`
1750 objects if ``model='pyrocko'`` is requested.
1752 See :py:meth:`iter_nuts` for details on time span matching.
1753 '''
1755 if model == 'pyrocko':
1756 return self._get_pyrocko_stations(obj, tmin, tmax, time, codes)
1757 elif model in ('squirrel', 'stationxml', 'stationxml+'):
1758 args = self._get_selection_args(
1759 STATION, obj, tmin, tmax, time, codes)
1761 nuts = sorted(
1762 self.iter_nuts('station', *args), key=lambda nut: nut.dkey)
1764 return [self.get_content(nut, model=model) for nut in nuts]
1765 else:
1766 raise ValueError('Invalid station model: %s' % model)
1768 @filldocs
1769 def get_channels(
1770 self, obj=None, tmin=None, tmax=None, time=None, codes=None,
1771 model='squirrel'):
1773 '''
1774 Get channels matching given constraints.
1776 %(query_args)s
1778 :returns:
1779 List of :py:class:`~pyrocko.squirrel.model.Channel` objects.
1781 See :py:meth:`iter_nuts` for details on time span matching.
1782 '''
1784 args = self._get_selection_args(
1785 CHANNEL, obj, tmin, tmax, time, codes)
1787 nuts = sorted(
1788 self.iter_nuts('channel', *args), key=lambda nut: nut.dkey)
1790 return [self.get_content(nut, model=model) for nut in nuts]
1792 @filldocs
1793 def get_sensors(
1794 self, obj=None, tmin=None, tmax=None, time=None, codes=None):
1796 '''
1797 Get sensors matching given constraints.
1799 %(query_args)s
1801 :returns:
1802 List of :py:class:`~pyrocko.squirrel.model.Sensor` objects.
1804 See :py:meth:`iter_nuts` for details on time span matching.
1805 '''
1807 tmin, tmax, codes = self._get_selection_args(
1808 CHANNEL, obj, tmin, tmax, time, codes)
1810 if codes is not None:
1811 codes = codes_patterns_list(
1812 (entry.replace(channel=entry.channel[:-1] + '?')
1813 if entry.channel != '*' else entry)
1814 for entry in codes)
1816 nuts = sorted(
1817 self.iter_nuts(
1818 'channel', tmin, tmax, codes), key=lambda nut: nut.dkey)
1820 return [
1821 sensor for sensor in model.Sensor.from_channels(
1822 self.get_content(nut) for nut in nuts)
1823 if match_time_span(tmin, tmax, sensor)]
1825 @filldocs
1826 def get_responses(
1827 self, obj=None, tmin=None, tmax=None, time=None, codes=None,
1828 model='squirrel'):
1830 '''
1831 Get instrument responses matching given constraints.
1833 %(query_args)s
1835 :returns:
1836 List of :py:class:`~pyrocko.squirrel.model.Response` objects.
1838 See :py:meth:`iter_nuts` for details on time span matching.
1839 '''
1841 args = self._get_selection_args(
1842 RESPONSE, obj, tmin, tmax, time, codes)
1844 nuts = sorted(
1845 self.iter_nuts('response', *args), key=lambda nut: nut.dkey)
1847 return [self.get_content(nut, model=model) for nut in nuts]
1849 @filldocs
1850 def get_response(
1851 self, obj=None, tmin=None, tmax=None, time=None, codes=None,
1852 model='squirrel'):
1854 '''
1855 Get instrument response matching given constraints.
1857 %(query_args)s
1859 :returns:
1860 :py:class:`~pyrocko.squirrel.model.Response` object.
1862 Same as :py:meth:`get_responses` but returning exactly one response.
1863 Raises :py:exc:`~pyrocko.squirrel.error.NotAvailable` if zero or more
1864 than one is available.
1866 See :py:meth:`iter_nuts` for details on time span matching.
1867 '''
1869 if model == 'stationxml':
1870 model_ = 'stationxml+'
1871 else:
1872 model_ = model
1874 responses = self.get_responses(
1875 obj, tmin, tmax, time, codes, model=model_)
1876 if len(responses) == 0:
1877 raise error.NotAvailable(
1878 'No instrument response available (%s).'
1879 % self._get_selection_args_str(
1880 RESPONSE, obj, tmin, tmax, time, codes))
1882 elif len(responses) > 1:
1883 if model_ == 'squirrel':
1884 resps_sq = responses
1885 elif model_ == 'stationxml+':
1886 resps_sq = [resp[0] for resp in responses]
1887 else:
1888 raise ValueError('Invalid response model: %s' % model)
1890 rinfo = ':\n' + '\n'.join(
1891 ' ' + resp.summary for resp in resps_sq)
1893 raise error.NotAvailable(
1894 'Multiple instrument responses matching given constraints '
1895 '(%s)%s' % (
1896 self._get_selection_args_str(
1897 RESPONSE, obj, tmin, tmax, time, codes), rinfo))
1899 if model == 'stationxml':
1900 return responses[0][1]
1901 else:
1902 return responses[0]
1904 @filldocs
1905 def get_events(
1906 self, obj=None, tmin=None, tmax=None, time=None, codes=None):
1908 '''
1909 Get events matching given constraints.
1911 %(query_args)s
1913 :returns:
1914 List of :py:class:`~pyrocko.model.event.Event` objects.
1916 See :py:meth:`iter_nuts` for details on time span matching.
1917 '''
1919 args = self._get_selection_args(EVENT, obj, tmin, tmax, time, codes)
1920 nuts = sorted(
1921 self.iter_nuts('event', *args), key=lambda nut: nut.dkey)
1923 return [self.get_content(nut) for nut in nuts]
1925 def _redeem_promises(self, *args, order_only=False):
1927 def split_promise(order):
1928 self._split_nuts(
1929 'waveform_promise',
1930 order.tmin, order.tmax,
1931 codes=order.codes,
1932 path=order.source_id)
1934 tmin, tmax, _ = args
1936 waveforms = list(self.iter_nuts('waveform', *args))
1937 promises = list(self.iter_nuts('waveform_promise', *args))
1939 codes_to_avail = defaultdict(list)
1940 for nut in waveforms:
1941 codes_to_avail[nut.codes].append((nut.tmin, nut.tmax))
1943 def tts(x):
1944 if isinstance(x, tuple):
1945 return tuple(tts(e) for e in x)
1946 elif isinstance(x, list):
1947 return list(tts(e) for e in x)
1948 else:
1949 return util.time_to_str(x)
1951 orders = []
1952 for promise in promises:
1953 waveforms_avail = codes_to_avail[promise.codes]
1954 for block_tmin, block_tmax in blocks(
1955 max(tmin, promise.tmin),
1956 min(tmax, promise.tmax),
1957 promise.deltat):
1959 orders.append(
1960 WaveformOrder(
1961 source_id=promise.file_path,
1962 codes=promise.codes,
1963 tmin=block_tmin,
1964 tmax=block_tmax,
1965 deltat=promise.deltat,
1966 gaps=gaps(waveforms_avail, block_tmin, block_tmax)))
1968 orders_noop, orders = lpick(lambda order: order.gaps, orders)
1970 order_keys_noop = set(order_key(order) for order in orders_noop)
1971 if len(order_keys_noop) != 0 or len(orders_noop) != 0:
1972 logger.info(
1973 'Waveform orders already satisified with cached/local data: '
1974 '%i (%i)' % (len(order_keys_noop), len(orders_noop)))
1976 for order in orders_noop:
1977 split_promise(order)
1979 if order_only:
1980 if orders:
1981 self._pending_orders.extend(orders)
1982 logger.info(
1983 'Enqueuing %i waveform order%s.'
1984 % len_plural(orders))
1985 return
1986 else:
1987 if self._pending_orders:
1988 orders.extend(self._pending_orders)
1989 logger.info(
1990 'Adding %i previously enqueued order%s.'
1991 % len_plural(self._pending_orders))
1993 self._pending_orders = []
1995 source_ids = []
1996 sources = {}
1997 for source in self._sources:
1998 if isinstance(source, fdsn.FDSNSource):
1999 source_ids.append(source._source_id)
2000 sources[source._source_id] = source
2002 source_priority = dict(
2003 (source_id, i) for (i, source_id) in enumerate(source_ids))
2005 order_groups = defaultdict(list)
2006 for order in orders:
2007 order_groups[order_key(order)].append(order)
2009 for k, order_group in order_groups.items():
2010 order_group.sort(
2011 key=lambda order: source_priority[order.source_id])
2013 n_order_groups = len(order_groups)
2015 if len(order_groups) != 0 or len(orders) != 0:
2016 logger.info(
2017 'Waveform orders standing for download: %i (%i)'
2018 % (len(order_groups), len(orders)))
2020 task = make_task('Waveform orders processed', n_order_groups)
2021 else:
2022 task = None
2024 def release_order_group(order):
2025 okey = order_key(order)
2026 for followup in order_groups[okey]:
2027 split_promise(followup)
2029 del order_groups[okey]
2031 if task:
2032 task.update(n_order_groups - len(order_groups))
2034 def noop(order):
2035 pass
2037 def success(order):
2038 release_order_group(order)
2039 split_promise(order)
2041 def batch_add(paths):
2042 self.add(paths)
2044 calls = queue.Queue()
2046 def enqueue(f):
2047 def wrapper(*args):
2048 calls.put((f, args))
2050 return wrapper
2052 while order_groups:
2054 orders_now = []
2055 empty = []
2056 for k, order_group in order_groups.items():
2057 try:
2058 orders_now.append(order_group.pop(0))
2059 except IndexError:
2060 empty.append(k)
2062 for k in empty:
2063 del order_groups[k]
2065 by_source_id = defaultdict(list)
2066 for order in orders_now:
2067 by_source_id[order.source_id].append(order)
2069 threads = []
2070 for source_id in by_source_id:
2071 def download():
2072 try:
2073 sources[source_id].download_waveforms(
2074 by_source_id[source_id],
2075 success=enqueue(success),
2076 error_permanent=enqueue(split_promise),
2077 error_temporary=noop,
2078 batch_add=enqueue(batch_add))
2080 finally:
2081 calls.put(None)
2083 thread = threading.Thread(target=download)
2084 thread.start()
2085 threads.append(thread)
2087 ndone = 0
2088 while ndone < len(threads):
2089 ret = calls.get()
2090 if ret is None:
2091 ndone += 1
2092 else:
2093 ret[0](*ret[1])
2095 for thread in threads:
2096 thread.join()
2098 if task:
2099 task.update(n_order_groups - len(order_groups))
2101 if task:
2102 task.done()
2104 @filldocs
2105 def get_waveform_nuts(
2106 self, obj=None, tmin=None, tmax=None, time=None, codes=None,
2107 order_only=False):
2109 '''
2110 Get waveform content entities matching given constraints.
2112 %(query_args)s
2114 Like :py:meth:`get_nuts` with ``kind='waveform'`` but additionally
2115 resolves matching waveform promises (downloads waveforms from remote
2116 sources).
2118 See :py:meth:`iter_nuts` for details on time span matching.
2119 '''
2121 args = self._get_selection_args(WAVEFORM, obj, tmin, tmax, time, codes)
2122 self._redeem_promises(*args, order_only=order_only)
2123 return sorted(
2124 self.iter_nuts('waveform', *args), key=lambda nut: nut.dkey)
2126 @filldocs
2127 def have_waveforms(
2128 self, obj=None, tmin=None, tmax=None, time=None, codes=None):
2130 '''
2131 Check if any waveforms or waveform promises are available for given
2132 constraints.
2134 %(query_args)s
2135 '''
2137 args = self._get_selection_args(WAVEFORM, obj, tmin, tmax, time, codes)
2138 return bool(list(
2139 self.iter_nuts('waveform', *args, limit=1))) \
2140 or bool(list(
2141 self.iter_nuts('waveform_promise', *args, limit=1)))
2143 @filldocs
2144 def get_waveforms(
2145 self, obj=None, tmin=None, tmax=None, time=None, codes=None,
2146 uncut=False, want_incomplete=True, degap=True, maxgap=5,
2147 maxlap=None, snap=None, include_last=False, load_data=True,
2148 accessor_id='default', operator_params=None, order_only=False):
2150 '''
2151 Get waveforms matching given constraints.
2153 %(query_args)s
2155 :param uncut:
2156 Set to ``True``, to disable cutting traces to [``tmin``, ``tmax``]
2157 and to disable degapping/deoverlapping. Returns untouched traces as
2158 they are read from file segment. File segments are always read in
2159 their entirety.
2160 :type uncut:
2161 bool
2163 :param want_incomplete:
2164 If ``True``, gappy/incomplete traces are included in the result.
2165 :type want_incomplete:
2166 bool
2168 :param degap:
2169 If ``True``, connect traces and remove gaps and overlaps.
2170 :type degap:
2171 bool
2173 :param maxgap:
2174 Maximum gap size in samples which is filled with interpolated
2175 samples when ``degap`` is ``True``.
2176 :type maxgap:
2177 int
2179 :param maxlap:
2180 Maximum overlap size in samples which is removed when ``degap`` is
2181 ``True``.
2182 :type maxlap:
2183 int
2185 :param snap:
2186 Rounding functions used when computing sample index from time
2187 instance, for trace start and trace end, respectively. By default,
2188 ``(round, round)`` is used.
2189 :type snap:
2190 tuple of 2 callables
2192 :param include_last:
2193 If ``True``, add one more sample to the returned traces (the sample
2194 which would be the first sample of a query with ``tmin`` set to the
2195 current value of ``tmax``).
2196 :type include_last:
2197 bool
2199 :param load_data:
2200 If ``True``, waveform data samples are read from files (or cache).
2201 If ``False``, meta-information-only traces are returned (dummy
2202 traces with no data samples).
2203 :type load_data:
2204 bool
2206 :param accessor_id:
2207 Name of consumer on who's behalf data is accessed. Used in cache
2208 management (see :py:mod:`~pyrocko.squirrel.cache`). Used as a key
2209 to distinguish different points of extraction for the decision of
2210 when to release cached waveform data. Should be used when data is
2211 alternately extracted from more than one region / selection.
2212 :type accessor_id:
2213 str
2215 See :py:meth:`iter_nuts` for details on time span matching.
2217 Loaded data is kept in memory (at least) until
2218 :py:meth:`clear_accessor` has been called or
2219 :py:meth:`advance_accessor` has been called two consecutive times
2220 without data being accessed between the two calls (by this accessor).
2221 Data may still be further kept in the memory cache if held alive by
2222 consumers with a different ``accessor_id``.
2223 '''
2225 tmin, tmax, codes = self._get_selection_args(
2226 WAVEFORM, obj, tmin, tmax, time, codes)
2228 self_tmin, self_tmax = self.get_time_span(
2229 ['waveform', 'waveform_promise'])
2231 if None in (self_tmin, self_tmax):
2232 logger.warning(
2233 'No waveforms available.')
2234 return []
2236 tmin = tmin if tmin is not None else self_tmin
2237 tmax = tmax if tmax is not None else self_tmax
2239 if codes is not None and len(codes) == 1:
2240 # TODO: fix for multiple / mixed codes
2241 operator = self.get_operator(codes[0])
2242 if operator is not None:
2243 return operator.get_waveforms(
2244 self, codes[0],
2245 tmin=tmin, tmax=tmax,
2246 uncut=uncut, want_incomplete=want_incomplete, degap=degap,
2247 maxgap=maxgap, maxlap=maxlap, snap=snap,
2248 include_last=include_last, load_data=load_data,
2249 accessor_id=accessor_id, params=operator_params)
2251 nuts = self.get_waveform_nuts(
2252 obj, tmin, tmax, time, codes, order_only=order_only)
2254 if order_only:
2255 return []
2257 if load_data:
2258 traces = [
2259 self.get_content(nut, 'waveform', accessor_id) for nut in nuts]
2261 else:
2262 traces = [
2263 trace.Trace(**nut.trace_kwargs) for nut in nuts]
2265 if uncut:
2266 return traces
2268 if snap is None:
2269 snap = (round, round)
2271 chopped = []
2272 for tr in traces:
2273 if not load_data and tr.ydata is not None:
2274 tr = tr.copy(data=False)
2275 tr.ydata = None
2277 try:
2278 chopped.append(tr.chop(
2279 tmin, tmax,
2280 inplace=False,
2281 snap=snap,
2282 include_last=include_last))
2284 except trace.NoData:
2285 pass
2287 processed = self._process_chopped(
2288 chopped, degap, maxgap, maxlap, want_incomplete, tmin, tmax)
2290 return processed
2292 @filldocs
2293 def chopper_waveforms(
2294 self, obj=None, tmin=None, tmax=None, time=None, codes=None,
2295 tinc=None, tpad=0.,
2296 want_incomplete=True, snap_window=False,
2297 degap=True, maxgap=5, maxlap=None,
2298 snap=None, include_last=False, load_data=True,
2299 accessor_id=None, clear_accessor=True, operator_params=None,
2300 grouping=None):
2302 '''
2303 Iterate window-wise over waveform archive.
2305 %(query_args)s
2307 :param tinc:
2308 Time increment (window shift time) (default uses ``tmax-tmin``).
2309 :type tinc:
2310 timestamp
2312 :param tpad:
2313 Padding time appended on either side of the data window (window
2314 overlap is ``2*tpad``).
2315 :type tpad:
2316 timestamp
2318 :param want_incomplete:
2319 If ``True``, gappy/incomplete traces are included in the result.
2320 :type want_incomplete:
2321 bool
2323 :param snap_window:
2324 If ``True``, start time windows at multiples of tinc with respect
2325 to system time zero.
2326 :type snap_window:
2327 bool
2329 :param degap:
2330 If ``True``, connect traces and remove gaps and overlaps.
2331 :type degap:
2332 bool
2334 :param maxgap:
2335 Maximum gap size in samples which is filled with interpolated
2336 samples when ``degap`` is ``True``.
2337 :type maxgap:
2338 int
2340 :param maxlap:
2341 Maximum overlap size in samples which is removed when ``degap`` is
2342 ``True``.
2343 :type maxlap:
2344 int
2346 :param snap:
2347 Rounding functions used when computing sample index from time
2348 instance, for trace start and trace end, respectively. By default,
2349 ``(round, round)`` is used.
2350 :type snap:
2351 tuple of 2 callables
2353 :param include_last:
2354 If ``True``, add one more sample to the returned traces (the sample
2355 which would be the first sample of a query with ``tmin`` set to the
2356 current value of ``tmax``).
2357 :type include_last:
2358 bool
2360 :param load_data:
2361 If ``True``, waveform data samples are read from files (or cache).
2362 If ``False``, meta-information-only traces are returned (dummy
2363 traces with no data samples).
2364 :type load_data:
2365 bool
2367 :param accessor_id:
2368 Name of consumer on who's behalf data is accessed. Used in cache
2369 management (see :py:mod:`~pyrocko.squirrel.cache`). Used as a key
2370 to distinguish different points of extraction for the decision of
2371 when to release cached waveform data. Should be used when data is
2372 alternately extracted from more than one region / selection.
2373 :type accessor_id:
2374 str
2376 :param clear_accessor:
2377 If ``True`` (default), :py:meth:`clear_accessor` is called when the
2378 chopper finishes. Set to ``False`` to keep loaded waveforms in
2379 memory when the generator returns.
2380 :type clear_accessor:
2381 bool
2383 :param grouping:
2384 By default, traversal over the data is over time and all matching
2385 traces of a time window are yielded. Using this option, it is
2386 possible to traverse the data first by group (e.g. station or
2387 network) and second by time. This can reduce the number of traces
2388 in each batch and thus reduce the memory footprint of the process.
2389 :type grouping:
2390 :py:class:`~pyrocko.squirrel.operator.Grouping`
2392 :yields:
2393 A list of :py:class:`~pyrocko.trace.Trace` objects for every
2394 extracted time window.
2396 See :py:meth:`iter_nuts` for details on time span matching.
2397 '''
2399 tmin, tmax, codes = self._get_selection_args(
2400 WAVEFORM, obj, tmin, tmax, time, codes)
2402 self_tmin, self_tmax = self.get_time_span(
2403 ['waveform', 'waveform_promise'])
2405 if None in (self_tmin, self_tmax):
2406 logger.warning(
2407 'Content has undefined time span. No waveforms and no '
2408 'waveform promises?')
2409 return
2411 if snap_window and tinc is not None:
2412 tmin = tmin if tmin is not None else self_tmin
2413 tmax = tmax if tmax is not None else self_tmax
2414 tmin = math.floor(tmin / tinc) * tinc
2415 tmax = math.ceil(tmax / tinc) * tinc
2416 else:
2417 tmin = tmin if tmin is not None else self_tmin + tpad
2418 tmax = tmax if tmax is not None else self_tmax - tpad
2420 tinc = tinc if tinc is not None else tmax - tmin
2422 try:
2423 if accessor_id is None:
2424 accessor_id = 'chopper%i' % self._n_choppers_active
2426 self._n_choppers_active += 1
2428 eps = tinc * 1e-6
2429 if tinc != 0.0:
2430 nwin = int(((tmax - eps) - tmin) / tinc) + 1
2431 else:
2432 nwin = 1
2434 if grouping is None:
2435 codes_list = [codes]
2436 else:
2437 operator = Operator(
2438 filtering=CodesPatternFiltering(codes=codes),
2439 grouping=grouping)
2441 available = set(self.get_codes(kind='waveform'))
2442 available.update(self.get_codes(kind='waveform_promise'))
2443 operator.update_mappings(sorted(available))
2445 codes_list = [
2446 codes_patterns_list(scl)
2447 for scl in operator.iter_in_codes()]
2449 ngroups = len(codes_list)
2450 for igroup, scl in enumerate(codes_list):
2451 for iwin in range(nwin):
2452 wmin, wmax = tmin+iwin*tinc, min(tmin+(iwin+1)*tinc, tmax)
2454 chopped = self.get_waveforms(
2455 tmin=wmin-tpad,
2456 tmax=wmax+tpad,
2457 codes=scl,
2458 snap=snap,
2459 include_last=include_last,
2460 load_data=load_data,
2461 want_incomplete=want_incomplete,
2462 degap=degap,
2463 maxgap=maxgap,
2464 maxlap=maxlap,
2465 accessor_id=accessor_id,
2466 operator_params=operator_params)
2468 self.advance_accessor(accessor_id)
2470 yield Batch(
2471 tmin=wmin,
2472 tmax=wmax,
2473 i=iwin,
2474 n=nwin,
2475 igroup=igroup,
2476 ngroups=ngroups,
2477 traces=chopped)
2479 finally:
2480 self._n_choppers_active -= 1
2481 if clear_accessor:
2482 self.clear_accessor(accessor_id, 'waveform')
2484 def _process_chopped(
2485 self, chopped, degap, maxgap, maxlap, want_incomplete, tmin, tmax):
2487 chopped.sort(key=lambda a: a.full_id)
2488 if degap:
2489 chopped = trace.degapper(chopped, maxgap=maxgap, maxlap=maxlap)
2491 if not want_incomplete:
2492 chopped_weeded = []
2493 for tr in chopped:
2494 emin = tr.tmin - tmin
2495 emax = tr.tmax + tr.deltat - tmax
2496 if (abs(emin) <= 0.5*tr.deltat and abs(emax) <= 0.5*tr.deltat):
2497 chopped_weeded.append(tr)
2499 elif degap:
2500 if (0. < emin <= 5. * tr.deltat
2501 and -5. * tr.deltat <= emax < 0.):
2503 tr.extend(tmin, tmax-tr.deltat, fillmethod='repeat')
2504 chopped_weeded.append(tr)
2506 chopped = chopped_weeded
2508 return chopped
2510 def _get_pyrocko_stations(
2511 self, obj=None, tmin=None, tmax=None, time=None, codes=None):
2513 from pyrocko import model as pmodel
2515 if codes is not None:
2516 codes = codes_patterns_for_kind(STATION, codes)
2518 by_nsl = defaultdict(lambda: (list(), list()))
2519 for station in self.get_stations(obj, tmin, tmax, time, codes):
2520 sargs = station._get_pyrocko_station_args()
2521 by_nsl[station.codes.nsl][0].append(sargs)
2523 if codes is not None:
2524 codes = [model.CodesNSLCE(c) for c in codes]
2526 for channel in self.get_channels(obj, tmin, tmax, time, codes):
2527 sargs = channel._get_pyrocko_station_args()
2528 sargs_list, channels_list = by_nsl[channel.codes.nsl]
2529 sargs_list.append(sargs)
2530 channels_list.append(channel)
2532 pstations = []
2533 nsls = list(by_nsl.keys())
2534 nsls.sort()
2535 for nsl in nsls:
2536 sargs_list, channels_list = by_nsl[nsl]
2537 sargs = util.consistency_merge(
2538 [('',) + x for x in sargs_list])
2540 by_c = defaultdict(list)
2541 for ch in channels_list:
2542 by_c[ch.codes.channel].append(ch._get_pyrocko_channel_args())
2544 chas = list(by_c.keys())
2545 chas.sort()
2546 pchannels = []
2547 for cha in chas:
2548 list_of_cargs = by_c[cha]
2549 cargs = util.consistency_merge(
2550 [('',) + x for x in list_of_cargs])
2551 pchannels.append(pmodel.Channel(*cargs))
2553 pstations.append(
2554 pmodel.Station(*sargs, channels=pchannels))
2556 return pstations
2558 @property
2559 def pile(self):
2561 '''
2562 Emulates the older :py:class:`pyrocko.pile.Pile` interface.
2564 This property exposes a :py:class:`pyrocko.squirrel.pile.Pile` object,
2565 which emulates most of the older :py:class:`pyrocko.pile.Pile` methods
2566 but uses the fluffy power of the Squirrel under the hood.
2568 This interface can be used as a drop-in replacement for piles which are
2569 used in existing scripts and programs for efficient waveform data
2570 access. The Squirrel-based pile scales better for large datasets. Newer
2571 scripts should use Squirrel's native methods to avoid the emulation
2572 overhead.
2573 '''
2574 from . import pile
2576 if self._pile is None:
2577 self._pile = pile.Pile(self)
2579 return self._pile
2581 def snuffle(self):
2582 '''
2583 Look at dataset in Snuffler.
2584 '''
2585 self.pile.snuffle()
2587 def _gather_codes_keys(self, kind, gather, selector):
2588 return set(
2589 gather(codes)
2590 for codes in self.iter_codes(kind)
2591 if selector is None or selector(codes))
2593 def __str__(self):
2594 return str(self.get_stats())
2596 def get_coverage(
2597 self, kind, tmin=None, tmax=None, codes=None, limit=None):
2599 '''
2600 Get coverage information.
2602 Get information about strips of gapless data coverage.
2604 :param kind:
2605 Content kind to be queried.
2606 :type kind:
2607 str
2609 :param tmin:
2610 Start time of query interval.
2611 :type tmin:
2612 timestamp
2614 :param tmax:
2615 End time of query interval.
2616 :type tmax:
2617 timestamp
2619 :param codes:
2620 If given, restrict query to given content codes patterns.
2621 :type codes:
2622 :py:class:`list` of :py:class:`~pyrocko.squirrel.model.Codes`
2623 objects appropriate for the queried content type, or anything which
2624 can be converted to such objects.
2626 :param limit:
2627 Limit query to return only up to a given maximum number of entries
2628 per matching time series (without setting this option, very gappy
2629 data could cause the query to execute for a very long time).
2630 :type limit:
2631 int
2633 :returns:
2634 Information about time spans covered by the requested time series
2635 data.
2636 :rtype:
2637 :py:class:`list` of :py:class:`Coverage` objects
2638 '''
2640 tmin_seconds, tmin_offset = model.tsplit(tmin)
2641 tmax_seconds, tmax_offset = model.tsplit(tmax)
2642 kind_id = to_kind_id(kind)
2644 codes_info = list(self._iter_codes_info(kind=kind))
2646 kdata_all = []
2647 if codes is None:
2648 for _, codes_entry, deltat, kind_codes_id, _ in codes_info:
2649 kdata_all.append(
2650 (codes_entry, kind_codes_id, codes_entry, deltat))
2652 else:
2653 for codes_entry in codes:
2654 pattern = to_codes(kind_id, codes_entry)
2655 for _, codes_entry, deltat, kind_codes_id, _ in codes_info:
2656 if model.match_codes(pattern, codes_entry):
2657 kdata_all.append(
2658 (pattern, kind_codes_id, codes_entry, deltat))
2660 kind_codes_ids = [x[1] for x in kdata_all]
2662 counts_at_tmin = {}
2663 if tmin is not None:
2664 for nut in self.iter_nuts(
2665 kind, tmin, tmin, kind_codes_ids=kind_codes_ids):
2667 k = nut.codes, nut.deltat
2668 if k not in counts_at_tmin:
2669 counts_at_tmin[k] = 0
2671 counts_at_tmin[k] += 1
2673 coverages = []
2674 for pattern, kind_codes_id, codes_entry, deltat in kdata_all:
2675 entry = [pattern, codes_entry, deltat, None, None, []]
2676 for i, order in [(0, 'ASC'), (1, 'DESC')]:
2677 sql = self._sql('''
2678 SELECT
2679 time_seconds,
2680 time_offset
2681 FROM %(db)s.%(coverage)s
2682 WHERE
2683 kind_codes_id == ?
2684 ORDER BY
2685 kind_codes_id ''' + order + ''',
2686 time_seconds ''' + order + ''',
2687 time_offset ''' + order + '''
2688 LIMIT 1
2689 ''')
2691 for row in self._conn.execute(sql, [kind_codes_id]):
2692 entry[3+i] = model.tjoin(row[0], row[1])
2694 if None in entry[3:5]:
2695 continue
2697 args = [kind_codes_id]
2699 sql_time = ''
2700 if tmin is not None:
2701 # intentionally < because (== tmin) is queried from nuts
2702 sql_time += ' AND ( ? < time_seconds ' \
2703 'OR ( ? == time_seconds AND ? < time_offset ) ) '
2704 args.extend([tmin_seconds, tmin_seconds, tmin_offset])
2706 if tmax is not None:
2707 sql_time += ' AND ( time_seconds < ? ' \
2708 'OR ( ? == time_seconds AND time_offset <= ? ) ) '
2709 args.extend([tmax_seconds, tmax_seconds, tmax_offset])
2711 sql_limit = ''
2712 if limit is not None:
2713 sql_limit = ' LIMIT ?'
2714 args.append(limit)
2716 sql = self._sql('''
2717 SELECT
2718 time_seconds,
2719 time_offset,
2720 step
2721 FROM %(db)s.%(coverage)s
2722 WHERE
2723 kind_codes_id == ?
2724 ''' + sql_time + '''
2725 ORDER BY
2726 kind_codes_id,
2727 time_seconds,
2728 time_offset
2729 ''' + sql_limit)
2731 rows = list(self._conn.execute(sql, args))
2733 if limit is not None and len(rows) == limit:
2734 entry[-1] = None
2735 else:
2736 counts = counts_at_tmin.get((codes_entry, deltat), 0)
2737 tlast = None
2738 if tmin is not None:
2739 entry[-1].append((tmin, counts))
2740 tlast = tmin
2742 for row in rows:
2743 t = model.tjoin(row[0], row[1])
2744 counts += row[2]
2745 entry[-1].append((t, counts))
2746 tlast = t
2748 if tmax is not None and (tlast is None or tlast != tmax):
2749 entry[-1].append((tmax, counts))
2751 coverages.append(model.Coverage.from_values(entry + [kind_id]))
2753 return coverages
2755 def get_stationxml(
2756 self, obj=None, tmin=None, tmax=None, time=None, codes=None,
2757 level='response'):
2759 '''
2760 Get station/channel/response metadata in StationXML representation.
2762 %(query_args)s
2764 :returns:
2765 :py:class:`~pyrocko.io.stationxml.FDSNStationXML` object.
2766 '''
2768 if level not in ('network', 'station', 'channel', 'response'):
2769 raise ValueError('Invalid level: %s' % level)
2771 tmin, tmax, codes = self._get_selection_args(
2772 CHANNEL, obj, tmin, tmax, time, codes)
2774 filtering = CodesPatternFiltering(codes=codes)
2776 nslcs = list(set(
2777 codes.nslc for codes in
2778 filtering.filter(self.get_codes(kind='channel'))))
2780 from pyrocko.io import stationxml as sx
2782 networks = []
2783 for net, stas in prefix_tree(nslcs):
2784 network = sx.Network(code=net)
2785 networks.append(network)
2787 if level not in ('station', 'channel', 'response'):
2788 continue
2790 for sta, locs in stas:
2791 stations = self.get_stations(
2792 tmin=tmin,
2793 tmax=tmax,
2794 codes=(net, sta, '*'),
2795 model='stationxml')
2797 errors = sx.check_overlaps(
2798 'Station', (net, sta), stations)
2800 if errors:
2801 raise sx.Inconsistencies(
2802 'Inconsistencies found:\n %s'
2803 % '\n '.join(errors))
2805 network.station_list.extend(stations)
2807 if level not in ('channel', 'response'):
2808 continue
2810 for loc, chas in locs:
2811 for cha, _ in chas:
2812 channels = self.get_channels(
2813 tmin=tmin,
2814 tmax=tmax,
2815 codes=(net, sta, loc, cha),
2816 model='stationxml')
2818 errors = sx.check_overlaps(
2819 'Channel', (net, sta, loc, cha), channels)
2821 if errors:
2822 raise sx.Inconsistencies(
2823 'Inconsistencies found:\n %s'
2824 % '\n '.join(errors))
2826 for channel in channels:
2827 station = sx.find_containing(stations, channel)
2828 if station is not None:
2829 station.channel_list.append(channel)
2830 else:
2831 raise sx.Inconsistencies(
2832 'No station or station epoch found for '
2833 'channel: %s' % '.'.join(
2834 (net, sta, loc, cha)))
2836 if level != 'response':
2837 continue
2839 response_sq, response_sx = self.get_response(
2840 codes=(net, sta, loc, cha),
2841 tmin=channel.start_date,
2842 tmax=channel.end_date,
2843 model='stationxml+')
2845 if not (
2846 sx.eq_open(
2847 channel.start_date, response_sq.tmin)
2848 and sx.eq_open(
2849 channel.end_date, response_sq.tmax)):
2851 raise sx.Inconsistencies(
2852 'Response time span does not match '
2853 'channel time span: %s' % '.'.join(
2854 (net, sta, loc, cha)))
2856 channel.response = response_sx
2858 return sx.FDSNStationXML(
2859 source='Generated by Pyrocko Squirrel.',
2860 network_list=networks)
2862 def add_operator(self, op):
2863 self._operators.append(op)
2865 def update_operator_mappings(self):
2866 available = self.get_codes(kind=('channel'))
2868 for operator in self._operators:
2869 operator.update_mappings(available, self._operator_registry)
2871 def iter_operator_mappings(self):
2872 for operator in self._operators:
2873 for in_codes, out_codes in operator.iter_mappings():
2874 yield operator, in_codes, out_codes
2876 def get_operator_mappings(self):
2877 return list(self.iter_operator_mappings())
2879 def get_operator(self, codes):
2880 try:
2881 return self._operator_registry[codes][0]
2882 except KeyError:
2883 return None
2885 def get_operator_group(self, codes):
2886 try:
2887 return self._operator_registry[codes]
2888 except KeyError:
2889 return None, (None, None, None)
2891 def iter_operator_codes(self):
2892 for _, _, out_codes in self.iter_operator_mappings():
2893 for codes in out_codes:
2894 yield codes
2896 def get_operator_codes(self):
2897 return list(self.iter_operator_codes())
2899 def print_tables(self, table_names=None, stream=None):
2900 '''
2901 Dump raw database tables in textual form (for debugging purposes).
2903 :param table_names:
2904 Names of tables to be dumped or ``None`` to dump all.
2905 :type table_names:
2906 :py:class:`list` of :py:class:`str`
2908 :param stream:
2909 Open file or ``None`` to dump to standard output.
2910 '''
2912 if stream is None:
2913 stream = sys.stdout
2915 if isinstance(table_names, str):
2916 table_names = [table_names]
2918 if table_names is None:
2919 table_names = [
2920 'selection_file_states',
2921 'selection_nuts',
2922 'selection_kind_codes_count',
2923 'files', 'nuts', 'kind_codes', 'kind_codes_count']
2925 m = {
2926 'selection_file_states': '%(db)s.%(file_states)s',
2927 'selection_nuts': '%(db)s.%(nuts)s',
2928 'selection_kind_codes_count': '%(db)s.%(kind_codes_count)s',
2929 'files': 'files',
2930 'nuts': 'nuts',
2931 'kind_codes': 'kind_codes',
2932 'kind_codes_count': 'kind_codes_count'}
2934 for table_name in table_names:
2935 self._database.print_table(
2936 m[table_name] % self._names, stream=stream)
2939class SquirrelStats(Object):
2940 '''
2941 Container to hold statistics about contents available from a Squirrel.
2943 See also :py:meth:`Squirrel.get_stats`.
2944 '''
2946 nfiles = Int.T(
2947 help='Number of files in selection.')
2948 nnuts = Int.T(
2949 help='Number of index nuts in selection.')
2950 codes = List.T(
2951 Tuple.T(content_t=String.T()),
2952 help='Available code sequences in selection, e.g. '
2953 '(agency, network, station, location) for stations nuts.')
2954 kinds = List.T(
2955 String.T(),
2956 help='Available content types in selection.')
2957 total_size = Int.T(
2958 help='Aggregated file size of files is selection.')
2959 counts = Dict.T(
2960 String.T(), Dict.T(Tuple.T(content_t=String.T()), Int.T()),
2961 help='Breakdown of how many nuts of any content type and code '
2962 'sequence are available in selection, ``counts[kind][codes]``.')
2963 time_spans = Dict.T(
2964 String.T(), Tuple.T(content_t=Timestamp.T()),
2965 help='Time spans by content type.')
2966 sources = List.T(
2967 String.T(),
2968 help='Descriptions of attached sources.')
2969 operators = List.T(
2970 String.T(),
2971 help='Descriptions of attached operators.')
2973 def __str__(self):
2974 kind_counts = dict(
2975 (kind, sum(self.counts[kind].values())) for kind in self.kinds)
2977 scodes = model.codes_to_str_abbreviated(self.codes)
2979 ssources = '<none>' if not self.sources else '\n' + '\n'.join(
2980 ' ' + s for s in self.sources)
2982 soperators = '<none>' if not self.operators else '\n' + '\n'.join(
2983 ' ' + s for s in self.operators)
2985 def stime(t):
2986 return util.tts(t) if t is not None and t not in (
2987 model.g_tmin, model.g_tmax) else '<none>'
2989 def stable(rows):
2990 ns = [max(len(w) for w in col) for col in zip(*rows)]
2991 return '\n'.join(
2992 ' '.join(w.ljust(n) for n, w in zip(ns, row))
2993 for row in rows)
2995 def indent(s):
2996 return '\n'.join(' '+line for line in s.splitlines())
2998 stspans = '<none>' if not self.kinds else '\n' + indent(stable([(
2999 kind + ':',
3000 str(kind_counts[kind]),
3001 stime(self.time_spans[kind][0]),
3002 '-',
3003 stime(self.time_spans[kind][1])) for kind in sorted(self.kinds)]))
3005 s = '''
3006Number of files: %i
3007Total size of known files: %s
3008Number of index nuts: %i
3009Available content kinds: %s
3010Available codes: %s
3011Sources: %s
3012Operators: %s''' % (
3013 self.nfiles,
3014 util.human_bytesize(self.total_size),
3015 self.nnuts,
3016 stspans, scodes, ssources, soperators)
3018 return s.lstrip()
3021__all__ = [
3022 'Squirrel',
3023 'SquirrelStats',
3024]